FoMRHI Comm 1987 C. J.

Coakley

ORPHARION and CITTERN FRET ANALYSIS and other ancient tunings, including
BAROQUE LUTES

This study began after completing the paper on Dowland’s lute tuning (Ref 1). During that
research some late topics were woven into the text, but this final work on wire-strung
instruments requires a separate piece. No attempt has been made to modify the former paper in
light of the new research, which should make new comparisons and conclusions clearer. This
later work would not have been possible without the former theory, and in order to follow the
analysis of orpharion and cittern tunings it will be helpful to read the ‘Dowland’ paper and the
earlier Ref 2. Some new relations have also been found between meantone and perfect tunings,
with possible ancient uses for positioning frets. The different tunings of citterns and renaissance
lutes have unexpectedly led to new ideas on baroque lute tunings. Readers interested only in this
area will need background from the CITTERN section. There is a short section on using a
simple tuning meter such as the Korg C40 for different temperaments, which should help the
practical application of meantones, or any temperament on other instruments.
Reflecting on the actual frets we see on modern lutes, and the complete absence for old lutes, I
recalled mentioning in Ref 2 that fixed brass frets on surviving old citterns etc could be valuable
evidence. Then a recent description of the orpharion made by John Rose (Ref 3) came to mind.
This was made in 1580, contemporary with John Dowland, and was reported to have a fretting
close to equal temperament (ET). The orpharion is interesting for analyses of early tuning
schemes, because it was specified for playing some of the lute pieces. Recordings of Dowland’s
complete lute solos have used the orpharion for about a dozen pieces. The sound has aspects of a
quiet expressive spinet or a clavichord. After working on the fretting schemes for lutes it has
been possible to make a complete analysis of Rose’s tuning.
The earlier booklet by Donald Gill (Ref 4) had a small photo of an orpharion with 9 courses,
made by Francis Palmer in 1617. This immediately suggested 1/6th comma meantone (SCM)
from the pattern of the frets, and I began to analyse from an enlargement. Later the ‘web’
revealed interest in wire-strung instruments (Ref 5), which are less prominent than lutes. This
focus is mainly for makers, and provides a drawing of the Palmer orpharion. On closer
examination there appeared to be deviations from an accurate tuning, with poor positions for
principal frets such as 5 and 7. This instrument has differently scaled strings such that the basses
are longer than the trebles by about 90mm. The scaling, similar to a harp or keyboard, permits a
good tone for the basses, and can also avoid breaking a long thin treble. The Palmer is favoured
over the Rose by modern makers, who may generally fit a modern exact SCM scheme
The cittern then came to notice and I could not neglect the many surviving specimens. In
contrast, there may be only two other orpharions, in Germany. And yes we have no bandoras,
which are effectively large orpharions with the fragile top string removed from a lute tuning.
The cittern is very different from lutes, but I had no practical acquaintance with it. However, I
could work out string parameters, and see that the low elasticity would produce a high
sensitivity of tuning to stretching at the frets. In addition, I am grateful to Peter Forrester (Ref 6)
for discussing at length many aspects of their history and tuning, derived from his making and
research. It was especially important to know whether the few specimens I analysed, and maybe
generalized from, were representative.
The present work is a detailed analysis with some new conclusions, so it is necessary to
summarize previous findings. I have half expected to find that someone had already made a full
study, but temperaments have not been a highly active area since Barbour, Dombois and Lindley
several decades ago (Refs 7 to 9). In 1983 Peter Forrester wrote that most old citterns he had
examined, and also the orpharions by Rose and Palmer, had frets positioned for 1/5th comma
meantone (Ref 10). Interestingly, this large degree of tempering is the further practical step
suggested in Ref 1 for a lute, on technical and musical grounds. The estimates were later refined
to SCM for citterns but often with a sharper first fret (Ref 11); near SCM for Palmer (Ref 5);
and for Rose a noticeable difference from ET (Ref 3). The method was a comparison of tracings
from the frets against strips marked with ideal fret positions. In the ancient fretting instructions
for lutes a strip was marked up with initial geometric constructions (eg Ref 1), which were then
transferred to the fingerboard. Final adjusted positions could also have been recorded on a strip,
copies made, and scaled for different string lengths. A strip allows a modern maker or player to
set up or check any temperament of choice, and it is advisable for a lute player to carry one
around.
Outline of analyses
The various instruments will be discussed in the following order: the Rose orpharion, several
citterns, and finally the Palmer orpharion. Rose’s instrument has equal length strings and is most
like the lute, so that it follows on well from the earlier Refs 1 and 2. It also presents least
problem for analysis, and there was no previous treatment. The citterns deserve some numerical
and historical priority, and begin to illustrate many intriguing problems of analysis. After the
citterns, difficulties with Palmer’s frets will not seem too unique, and this popular model will
round off comparisons with the lute.
There are both similarities and differences between assessing real fret positions rather than
theoretical fretting instructions. Some are obvious, but others are subtle and have important
consequences, so preliminary explanation is necessary. Most of the analysis will plot the
measurements of fret positions on the diagram devised in Refs 1 and 2. This showed fret
positions for all degrees of temper between Pythagorean (PYT) and 1/3rd comma meantone
(TCM). A key interest is the sensitivity of tuning and tempering to the position of each fret, and
vice versa. This can be seen clearly on the diagram, but is not available from one or two strips.
The fundamental relations follow from the sequence of equal fourths, , which are the basis of
meantone. Each fret can be treated mathematically using the equations involving , or with
sufficient accuracy from the tables in Ref 2. Many of the calculations are converted to a 600mm
string length, which is useful for comparing different instruments. It is also necessary to retain
the real dimensions, since the absolute accuracy required in positioning and fitting frets is also
important.
In Ref 1 the ancient theoretical schemes were quantified, their origins and intentions discussed,
and probable adjustments of the gut frets to a final tuning were indicated. Some features of the
present analysis are similar, but the big contrast is that we are now dealing with real frets on
instruments nearly 500 years old. With the marvellous evidence of these surviving fixed frets,
the tuning system is final, with limited possible adjustment. New factors are the precision of the
maker in setting up his reference strip, then transferring it to the fingerboard, and finally cutting
the slots and fitting the frets. These further steps could involve adjustments of an initial scheme,
but also introduce practical errors. We now need to consider how well tuned the frets would be,
in addition to deducing a temperament. Another complicating factor is the extreme sensitivity of
tuning to the stretching of the wire strings, and an ancient maker may have adjusted initial fret
positions to allow for this effect. A similar but much smaller stretching effect occurs with gut
lute strings. The theoretical lute schemes already needed large adjustments that would have
included this effect, so that no special consideration was necessary in Ref 1. In Ref 12 I devised
a treatment of the stretching effects but just gave some examples. The complete theory is given
below because the effect is much larger and of great importance in making early wire-strung
instruments.
By these methods it has been possible to analyse and compare several citterns, and the
orpharions by Rose and Palmer. Some detailed suggestions can be made on the intended tunings
and how the fixed frets relate to the written instructions of Gerle, Dowland et al, and other
methods. The ancients’ approach to tuning, positioning and fitting of frets was clearly more
original, and of greater fascination to us, than our modern exact selections ‘off the shelf’. At the
start of this work such conclusions would have been an unimaginable endpoint. There are
implications for modern attitudes to all 16th and 17th century music for fretted strings, beyond
simple classifications of fret tunings in cents. Some further bonuses have arisen. Analysis of the
Rose orpharion has brought to attention several intriguing sequences of frets. These may have
been ancient guides to setting fret positions. They also provide a way of deducing the tuning of
lutes etc, by a single glance and pattern recognition. Studying the differences between the
tunings of citterns and lutes has led to new general criteria for tuning any fretted strings; the role
of natural or just intonation; and most surprisingly tuning systems for baroque lutes. There are
lengthy sections below on all these related topics.

ORPHARION by ROSE
Rose is credited with inventing both the orpharion and bandora. His surviving orpharion has six
double courses of equal length strings and none of the angled features in pictures of other
orpharions and bandoras. The top course of an orpharion required very thin ‘steel’ wire of
considerable strength and when supplies decreased the orpharion declined, to less legendary
acclaim than its namesakes Orpheus and Arion. (Bandora may derive from ‘pandoura’, a form
of Greek lute, although lyres and kitharai were preferred.) An orpharion with long upper strings
would have been particularly vulnerable. The present orpharion is unusual and lute-like, and
maybe it was played by famous lutenists. At the very least its fixed brass frets should give some
valuable evidence on a tuning temperament in use for fretted instruments at that time. For this
purpose it is important that the string length, the elaborately decorated neck and the frets look
unaltered. However, the bridge is an inelegant replacement, but with the saddle probably in the
original position, and two pegs are later additions (Ref 3). The whole instrument is exquisite in
overall design and detail, and characteristic of English Tudor art, in contrast with the Italian and
Middle Eastern lute. It is fragile now, and was not fully strung in recent studies (Ref 6). Fifty
years ago Gill said it sounded like a psaltery, or a harpsichord plucked by hand (Ref 13). He also
noted wear from a right little finger, which suggests it was originally well enough in tune.
In Ref 3 the fret positions were checked, listed in detail, then compared with ET and said to be
slightly different. In the following full analysis of these fret positions, this valuable practical
data needs to be treated differently from the previous theoretical schemes. It is useful to quote
the raw data as:
Nut 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Bridge
mm 35.0 64.2 94.0 122.2 150.2 175.0 199.1 222.3 242.6 261.5 280.7 298.7 315.8 331.9 346.0 599.1
Fifteen frets are normal on these instruments, and citterns often had nineteen. This is a contrast
with the standard twelve for a lute, ten or less for a baroque model, and only seven on Gerle’s
early lute. Most of the work will concentrate on the first octave. Theoretically, the higher octave
should be a scaled half-size repeat, and frets 13 to 15 can be used as a check, or a source of
further clues. In this respect fret 13 is reasonable, with 315.8 – 298.7 = 17.1, which is close to
17.5 or half of 35.0. Fret 14 has 331.9 – 298.7 = 33.2, but this is somewhat larger than 32.1mm.
Fret 15 has 47.3, which is close to 47.0. Checking against the photo of the neck, fret 14 has
indeed been measured correctly, rather than perhaps a 330.9 that could resolve the matter.
First it is necessary to plot the fret positions on the diagram. Frets 1 to 12 are shown by
prominent circles with crosses. Initially fret 12 is taken as an accurate octave, for later review.
The doubled values for frets 13 to 15 are shown by squares on the first three frets. For later use,
the squares in brackets represent perfect fifths on frets 1 to 4; and the letter P indicates perfect
intervals. The solid dots represent a shift of the nut 2mm away from the frets, for later
discussion of wire strings and various types of error.
An immediate impression is that the scatter is far less than for the theoretical schemes in Ref 1.
Eight frets lie between SCM and ET, and six between ET and PYT. In greater detail, frets 1, 2,
4, 5, 8, 11, 13 are close to 1/8th comma meantone (ECM), but frets 3, 7 and the less crucial frets
9, 10, 14, 15 are nearer to PYT. The fretting is closest to ECM, and this would seem the best
choice for a modern copy of this instrument. This is very different and more precise than saying
that a theoretical scheme might be adjusted to ECM.
For the later analysis it is useful to list the nearest tempered system for each fret as:
Fret 1 2 3 4 5 6 7 8 9 10 11 12
ECM ECM PYT ECM (SCM) ET (PYT) SCM ET PYT ET Octave
/SCM /SCM
As seen on the diagram and discussed in Ref 1, frets 5 and 7 vary only slightly over all of the
tempered systems, so that extraordinarily high accuracy would be needed for them alone to
define a particular system. For fret 7 a position close to PYT could be satisfactory for other
systems. Definition of a precise ECM would require a very slight decrease from 199.1 to
198.5mm. The scaled values, for a 300mm fret 12 on a 600mm string, would be 200 and
199.3mm. Remarkably, fret 5 at 150.2 in SCM was placed significantly differently from the
PYT value of 149.4. These small differences are probably close to the accuracy of marking a
strip, sawing etc. As discussed in Ref 1, this degree of fine-tuning was probably beyond ancient
theory. For a lute, later adjustment on a simple PYT could have been made with the gut frets,
and the present case might have used some trial metal frets. One can conclude that frets 5 and 7
are reasonably well tuned, but might have been slightly better. These points would have been
unnecessary in discussing the initial theoretical schemes for lutes, but are very important for
analysing real fixed frets.
A related concern is the position of the two end points for the open strings. These are obvious in
a theoretical scheme, but for this real instrument the original positions for the nut and bridge
saddle may be less certain than for the frets, which were carefully set in a finely decorated neck.
The analysis so far indicates that the nut position is good, and any uncertainty in the
measurement was a small 0.25mm (Ref 6).
Frets 5 and 7 can provide a further guide that is independent of an accurate nut position. For a
600mm string the spacing between these two frets should be 50mm in PYT, slightly reduced to
49.0mm in ET, and 48.3mm in SCM, from Table 7 in Ref 2. The scaled spacing for Rose’s frets
is fine at 49.1mm, but later instruments will show much less consistency. It will be seen that
these two frets are best used to test whether an instrument has been made correctly and is in
tune, or perhaps suggest alternative nut positions. As was shown for the theoretical schemes,
frets 5 and 7 cannot be used to determine a temperament. For example, low accuracy or a simple
error with the strip or saw should not be interpreted as tempering way beyond PYT or TCM.
For the bridge position, it is seen that fret 12 for the octave is slightly closer to the nut than half
the open length of 299.5mm, by 0.8mm. For a lute with gut strings, a similar measure would be
needed to correct for stretch sharpening when pressing down to the frets. It will be seen later
that this indicates the action height and frets on this instrument were very well adjusted to
minimize potentially much larger effects from metal wire. A possible problem is that bridges
and saddles may have been poorly set or replaced. These complications will arise in the analyses
of citterns and Palmer’s orpharion, so we are fortunate to have the present consistent example.
Having outlined some important differences from the theoretical schemes, each fret can be
examined, with some attention to how the ancients may have set the positions. A plausible
method would involve theoretical schemes, then refinement by playing on some device or by
experience. Making scaled copies of a standard strip would have been a possible early method.
Fret 1 is close to ECM and a clear possibility is that the type of tuning desired by Gerle, and
even his actual method, were being followed. Ref 1 demonstrated how Gerle’s method for SCM
could have been used for intermediate temperings, since no other suitable fractions exist.
Dowland had not yet written his piece of 1610 based on Gerle, but Rose’s earlier use of this type
of tuning is consistent with its familiarity in England.
Fret 3, however, is close to PYT, with no indication of copying the clever part of Gerle’s
scheme. This foreshadows Dowland’s intriguing departure from Gerle, but uses the simple
Pythagorean minor third, which Ref 1 suggested as an easier and better alternative for Dowland.
Fret 2 is about 1/7th comma, fret 4 about ECM, and fret 5 is SCM. As seen in Ref 1, fret 2 was
generally taken as Pythagorean, although I devised a tempered variation on Gerle’s
constructions. In Rose’s case it can be seen that fret 2 is close to a midpoint of frets 1 and 3.
Many ancient tunings took fret 5 as a perfect fourth and a tempered construction would have
been much more difficult. It is likely that Rose’s fret 5 was adjusted purely by ear, possibly via
other strips, or the happy result of a small error. Gerle’s fret 4 was in ET and constructed as
between frets 3 and 5, while Dowland later produced some confusion with his perfect major
third in QCM. Rose’s fret 4 is also at the midpoint of his 3 and 5. A detailed study of how
midpoints are involved in Rose’s fret positions is given in the following section.
Fret 7 is the normal ancient perfect fifth, and fret 6 in ET is the midpoint of frets 5 and 7. This
simple construction was common to all the schemes analysed in Ref 1, and here it is the final
fixed position in ET. Modern players often have a problem in deciding between a more common
Db for the top string and the sharp alternative. The ancients took a simplifying view in theory,
and also in this practical case. However, this acceptable compromise does not place the entire
system in ET.
Fret 8 is SCM and could be an exact perfect fifth construction from fret 1. This method was
described by Dowland and several others, as discussed in Ref 1.
Fret 9 is near ET, and slightly sharper by about 0.6mm than a perfect fifth from fret 2.
Fret 10 is exactly PYT and flatter by only 0.3mm than a perfect fifth from fret 3.
Fret 11 is near ET, and an exact perfect fifth from fret 4.
These higher frets are not so crucial as the first seven. Although the tuning varies between SCM
and PYT the absolute deviations from ECM are not large, and could be understandable errors in
constructions, marking out, or making. However, the resulting small space between frets 9 and
10 looks peculiar in comparison with the large space between 10 and 11.
Fret 12 appears to be satisfactory, fret 13 is close to ECM like fret 1, and fret 15 is slightly
sharper than fret 3 and quite close to ET. As discussed above, fret 14 is effectively sharper than
fret 2 by about 1mm. It is possible that a simple initial construction in PYT was not adjusted like
fret 2. Overall, the arrangement above fret 7 appears a little more disorderly and scattered, both
in the analysis and visually. This could result from the ancients’ theoretical difficulty with
higher frets or their tendency to build a scale by successive intervals, as explained in Ref 1. Also
the absolute accuracy of position for defining a system needs to be about twice that for the lower
frets.
In summary, the best classification of Rose’s tuning would be ECM, or 1/8th comma meantone,
and detailed examination indicates that the most significant deviations are frets 3 and 10. Most
of the principal lower frets are ECM rather than SCM or ET. It is clear that Rose’s fretting does
not closely resemble a simple gradation of equal semitones, such as Galilei’s 18/17 rule, which
would be a very distinctive pattern. There are only slight features of the Pythagorean schemes,
Ganassi or Bermudo. Interestingly, the first five frets are very similar to the ECM scheme I
derived in Ref 1, but it is unlikely that the ancients had discovered the necessary fractions. It
was also shown that for the ancients to obtain an intermediate tempering between SCM and ET
they could have needed to modify Gerle’s scheme by ear and eye. Rose’s frets appear
characteristic of the tempered tuning which Gerle had written about 35 years earlier, and was
probably familiar to English musicians throughout Dowland’s life. In fact, Rose’s frets look like
an adjusted version of Dowland’s later written instructions (see Ref 1). Rose was primarily a
viol maker, so there may also be some relation to tuning schemes for viols, for which there may
be no surviving instructions. Naturally, all these inferences must be taken with the usual
cautionary provisos and nuancing of historical research. (Ideally one would like relevant
writings by Rose, and a good picture of Dowland’s fingerboard.)
Rose’s orpharion was evidently played and it would be useful to know how the tuning sounded.
This is the final interest here, rather than how theoretical initial fret positions might be adjusted.
Modern makers often indicate that tunings of old frets have defects, but I have not seen any
reference to an exact copy of Rose’s fretting. Modern instruments may be tuned to an exact
modern SCM or ET. Assessing the actual tuning by theoretical methods would need enormous
amounts of comparison, including deviations due to stretching at the frets. An experimental test
rig would be ideal, but a simpler initial test could start with movable lute frets. The largest
deviation would be expected to arise from Rose’s flat third fret, so I tested this on a lute tuned in
SCM. The result of a 2mm flattening on the fret spacing of 32mm did not sound too disastrous.
The size of the effect is about 6cents (100 x 2/32), and one can see that although most intervals
are worse, some such as minor thirds become more perfect. It is worth noting that this fret is the
most frequently used. A pale fingerboard can show a heavy use across all main courses, which
decreases in amount and extent towards frets 1 and 8.

FRET PATTERNS
All players are familiar with the smooth gradation of spacings between frets in ET. Others will
recognize, and perhaps be puzzled by, the alternate wide and small spacings of SCM frets.
These patterns were explained in Ref 2 and can be seen on the diagram. Fewer people may
encounter the pattern in ECM which is a milder version of SCM, and still very distinct from a
nearby ET. Although performers do not generally mention their temperaments, photos of ECM
accompany at least two prominent lute recordings. The analysis of Rose’s orpharion has
revealed some new patterns. These can be used to identify temperaments, and may even have
been used by the ancients when setting up fret positions.
The photos of frets in Ref 3 do not bring to mind an exact ECM, SCM or ET. Some spacings
seem equal, and ancient fretting instructions had used midpoints. This geometric construction is
very simple and can avoid otherwise difficult formulations. A good way of investigating Rose’s
frets is a drawing accompanied by fret spacings to the nearest mm. This method, which was also
used in Ref 1 to search for fractions, reveals the quantitative patterns more clearly than decimal
measurements from the nut. The unscaled spacings can be listed as:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12 12/13 13/14 14/15
35 | 29 30 | 28 28 | 25 24 | 23 | 20 19 | 19 18 || 17 16 | 14
mm
One can discern pairs of approximately equal spacings, as marked by vertical dashes.
For an exact SCM scaled to 600mm the pattern is:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12 12/13 13/14 14/15
36 | 28 32 | 25 29 | 27 21 | 24 | 19 22 | 17 19 || 18 | 14 16
There are no equal pairs, and the pattern of alternate wide and small spacings is best marked in
whole tones as:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
36 28 | 32 25 | 29 27 | 21 | 24 19 | 22 17 | 19 ||
| | || || | | ||
It is also helpful to mark off frets 5 and 7 to show the structure of the scale with the diatonic
semitones 4/5 and 11/12, and leave out the higher octave.
For ET the approximate scaled spacings are able to retain the graded pattern as:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
33 32 | 30 28 | 27 25 | 24 | 22 21 | 20 19 | 18 ||
| | || || | | ||
For ECM the pattern is
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
35 | 30 31 | 27 28 | 26 22 | 23 | 21 21 | 18 19 ||
| | || || | | ||
This shows pairs of almost equal spacings, but not so distinctly as Rose’s frets. Careful use of a
ruler on the diagram shows that equal spacings such as 1/2 and 2/3 do in fact lie on a horizontal
between ET and ECM.
It was interesting to see whether an exact calculation could be made, and also for other pairs of
spacings. From Ref 2, the distance of a fret from the nut can be expressed in terms of the
frequency ratio, and hence the effective value of the fourth, . This leads to a condition for
spacings 1/2 and 2/3 to be equal at a value of  given by the equation 7 = 22 + 4. One is not
usually happy to see general seventh order equations, but in this case we know  is not too
different from a perfect fourth. If the tempering is say x comma then  can be expressed as
(4/3)(81/80)x  (4/3) (1+ x/80). Eventually the solution is found as x  0.113, giving 1/(8.8)th or
about 1/9th comma.
More important than the precise value of , conditions for other equal spacings can be found.
The same method shows that 3/4 and 4/5 are also equal under the same condition (1) found
above.
It can be seen that over the whole range of temperaments from PYT to TCM, the spacings 4/5
and 5/6 are almost equal. The exact condition is 5 = 4 or x = -0.406. This occurs way beyond
PYT where the spacings are only exactly equal when they are zero. Rose’s frets do not show the
approximate equality because fret 6 is midway between 5 and 7, so that 5/6 and 6/7 are equal,
which is not true for the theoretical forms between ET and SCM. It is also seen that while his
lower frets are mainly ECM, there are also equal pairs, which in theory require the lesser 1/9th
comma. His strangely flat fret 3 reduces 2/3 and increases 3/4 to produce an equal pair, unlike
the exact ECM above.
Continuing with higher frets, condition (1) can be shown to determine equality of 6/7 with 7/8;
8/9 with 9/10; and 10/11 with 11/12. In summary, five separate pairs of spacings are equal for
exactly the same condition close to 1/9th comma meantone. In addition, the second pair is next to
another almost equal spacing, giving effectively an equal triplet. All this information is packed
inside the diagram, but it is not obvious and needs discovery and explanation.
All the equal spacings in the 1/9th comma scaled pattern can be written in whole mm as:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
34 | 30 30 | 27 27 27 | 23 23 | 20 20 | 19 19 ||
| | || || | | ||
A pattern of equal spacings could be used by a theorist or maker to construct initial fret
positions, or refine them. Although simple linear division, such as midpoints and trisection, have
no direct simple basis they can have a deeper theoretical justification. From the work of Rose,
Gerle et al, this may have been a stronger guide to the ancients, than for example a simple
graded pattern of exact ET, which may have been unknown or of little interest. Fret 2, for
example, is at the midpoint of 1 and 3, but would have been difficult to place in ECM by other
theoretical means. Similarly, fret 4 for the difficult major third is the midpoint of 3 and 5.
Dowland later used this midpoint, whereas Gerle had just placed the fret in between. The
ancients generally set fret 6 as the midpoint of Pythagorean frets 5 and 7. Rose has two further
midpoints, frets 9 and 11, which appear to be better tuned than the intermediate fret 10. Perhaps
he forgot to adjust some constructional scaffolding in the related frets 3 and 10.
Between PYT and ET there are some further interesting patterns. These can be outlined for
completeness, and may be relevant for earlier medieval tunings. In Ref 1 it was noted that later
musicians may have strayed into this area, and recognition of these patterns could prevent this.
For PYT the pattern is:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
30 36 | 27 32 | 25 23 | 27 | 20 24 | 18 21 | 16 ||
| | || || | | ||
The alternate large and small spacings are a reversal of SCM, and use of a ruler on the diagram
indicates that different pairs can be equal for tempering slightly less than ET. The first spacing
0/1 next to the nut can now equal 1/2, and so on. This occurs at a value of  given by the
equation 7 = 16 - 25. The solution is x  0.068, giving 1/(14.7)th comma, or about 1/15th
comma. This can be called condition (2) and the following pairs of spacings are equal for this
same value of : 0/1 and 1/2; 2/3 and 3/4, 5/6 and 6/7; 7/8 and 8/9; 9/10 and 10/11. In addition,
4/5 and 5/6 are almost equal, as on the other side of ET. This gives five separate equal pairs of
spacings, as for condition (1), and the third pair becomes an almost equal triplet.
The scaled spacings for this 1/15th comma are:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
33 33 | 29 29 | 25 25 25 | 22 22 | 20 20 | 17 ||
| | || || | | ||
The pattern now starts at the nut and is even more visually striking than for 1/9th comma or
Rose’s frets. The two new distinct patterns, on each side and close to ET, will be a surprise. A
reversed later use could have been accurate guides for obtaining an intermediate exact ET with
continuous gradation and no equalities. All these possible methods might be grouped as
Coakley3 similar to the additional constructions in Ref 1.
The lists of spacings clearly show the patterns, but it is not easy to ensure visual recognition.
The patterns seem to stand out best when displayed as bands with a width similar to the
spacings, as on a neck with little taper, like Rose’s. A large display of a complete sequence of
cases seems to lose the distinctive appearances. Readers may like to make their own diagrams. If
one is sure enough, then a single glance at a performers’s lute, even before playing, could
identify six distinctive tunings between PYT and SCM. This seems better than the photo and
ruler in Ref 1. However, patterns are only in tune if they have the correct quantitative relations.
As a simple guide, the different patterns can be expressed in terms of large, small, or equal
spacings - L, S, E.
ECM, SCM QCM: | L S | L S | L | L S | L S | L S | L |
th
1/9 comma: | L | E1 E1 | E2 E2 E2 | E3 E3 | E4 E4 | E5 E5 |
ET: | L decreasing to S |
1/15th comma | E1 E1 | E2 E2 | E3 E3 E3| E4 E4 | E5 E5 | S |
PYT | S L | S L | S | S L | S L | S L | S |
For future reference it is useful to list the scaled spacings for the highly tempered QCM as:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12
39 24 | 35 22 | 31 29 | 18 | 26 16 | 24 14 | 21 || 20 12 | 18 11
The contrast between large and small spacings is extreme, with variations of over 60% for
adjacent frets. The largest ‘small’ spacing of 24mm is barely the width of a finger, and the
smallest is less than half at 11mm. This is significant for a small cittern, to which we now turn.

CITTERNS
There are about 35 known surviving citterns, providing a large sample for tuning analysis. I
have looked at only four, determined mainly by easy availability of fret measurements or good
photos. These seem to be representative (Ref 6), but there will be no dogmatic conclusions, and
the method of analysis could be applied to other examples.
Background information on citterns can be found elsewhere, but a few comments will put the
present inquiry in context. Their bodies are flat and rather small, with fret 12 invariably just
above a small finial at the top of the body. This would have been a reference point in designing
and playing. The neck is slender and the pegbox is topped by a carved head. These features also
appeared on the orpharion and bandora. The wavy outline of their bodies may reflect the shell
lyre of Orpheus, or Arion’s encounter with a dolphin. The string length on citterns is generally
in the range 40 to 50cm, but some are up to 60 and down to 35cm.
The basic cittern had four double courses of equal length tuned to e, d, g, and b or a, often with
octave strings on the lower two courses. The various arrangements of fixed brass frets initially
seems confusing since several, such as fret 4, were missed out on earlier citterns. These so called
‘diatonic’ tunings were adequate for the music, and similarly some frets were of smaller width
and acted on fewer courses. Later frets were fully ‘chromatic’, and the analysis will treat this
general case. The positions of diatonic frets need not be inherently different, and the
nomenclature itself is examined below. Missing frets would also slightly reduce effort in
construction and problems with the action of the strings. The low elasticity of metal wire makes
the tuning very sensitive to the height of the frets and the strings. Features found in the fretting
patterns will require a full analysis of the large degree of stretch-sharpening. The range of the
open strings, from g to the octave on b, is just over half that for a six course lute or the
orpharion. This may reduce the problems on citterns, and using a unison 4th course narrows the
range to only an octave.
In order to analyse the frets, some comments are needed on the allowed types of tuning. For a
lute it was shown in Refs 1 and 2 that any consistent tuning should be some form of meantone.
On a renaissance lute two octaves encompass four fourths and a third, which provided a neat
demonstration of the syntonic comma. For any meantone using simple frets, the interval of a
fourth or fifth between successive strings generates identical notes, except for a single altered
enharmonic on fret 11. It will be seen that this is a key factor for any fretted string instrument.
The full implications of the following generalization of the lute may be novel. Since each
interval of a fourth, or fifth, introduces a further new note, it is best for open strings to be
separated by only a limited number of fifths. For the lute we have in effect F, C, G, D, A and
hence four successive fifths. For these tuning purposes the strings do not need to be adjacent,
but it will be seen later that the position of a course and its musical use are highly important
limiting factors.
Extending this to a cittern we have G, D, A, E for a French type, involving three successive
fifths. This can therefore have the maximum number of useful common enharmonics for four
strings, as the lute had for five strings plus a second G. For the Italian or English cittern the
sequence of open strings is G, D, (A) E, B. The tuning involves four successive fifths, as for the
lute, even though there is no A string. This can still have a consistent meantone, but with fewer
common enharmonics than the French type. To illustrate the principle more strongly, the same
tuning would hold for just two open strings, in G and B. One can also see why further or
alternative strings on a lute, such as E, B, F#, Bb etc, would not be heipful. A hypothetical case
with two strings in say F and B, which are six fifths apart, would have a meantone that is far
more consistent than any other type of keyboard or natural tuning could be for fretted notes.
However, while most naturals would be preserved there would be no common enharmonics. If
flats were chosen for the F string then the B string would have sharps only. This general analysis
could be extended to types of citterns with more courses. As shown below, this theory can also
lead to new conclusions on the tuning of later baroque lutes. This is a very unexpected result of
studying a different much earlier instrument. There is a detailed treatment in a later section.
Four real citterns are now analysed: an English one, two Italians and a Dutch one, but this is no
joke. The attempt at analysis raises difficult questions and, if the explanations are incomplete,
the ideas might assist those with more data.
English cittern
The first measurements I obtained came from the National Music Museum website, brought to
notice by Darryl Martin. The English cittern at Dakota has a string length of only 35cm. It is
thought by some to be the only known English cittern. However, it may be a converted gittern,
but with frets that appear to be authentic and representative (Ref 6). There are 17 chromatic
simple frets serving all four courses. The frets are not angled, which appears to be a general
feature of citterns, in common with the Rose orpharion. This has implications for possible
stretch sharpening effects, and also for attitudes of lute makers to frets on their instrument. The
positions of frets and the nut from the bridge were given in mm. A cittern bridge is movable, as
on a violin, with the strings attached to a ‘comb’ at the tail. An uncertainty in string length and
also the nut position is a potential problem for analysis, but useful for a maker. Nevertheless, we
can obtain reasonable fret positions from the nut, but only to a nearest mm, and these check with
the photos. A probable random error of plus or minus ½ mm needs to be born in mind
throughout the following discussions.
The fret positions are plotted on the diagram. As above, fret 12 is initially taken as an accurate
octave. Frets 1 to 12 are shown by circles with crosses; the relative values for frets 13 to 17 are
shown by squares plotted on the first five frets. For later use, the squares in brackets represent
perfect fifths on frets 1 to 4, and the letter P indicates perfect intervals.
As above, it is best to start with a list of the measured fret positions, in mm:
Nut 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Bridge
mm 22 36 55 69 86 101 113 126 138 150 161 171 182 190 198 205 213 (350)
For the following detailed analysis it is useful to show the fret positions that would be required
for an exact SCM. It is appropriate to show some finer detail of ½mm, without the obscuring
effect of decimals:
N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
20½ 36½ 55 69½ 86 101½ 113½ 127½ 138 150½ 160 171 181½ 189½ 198½ 205½ 214
This shows immediately that the main differences are a sharper, more tempered fret 1 and a
flatter fret 8 on the cittern.
The temperament of each fret was also deduced and the approximate values can be listed as:
Fret 1 2 3 4 5 6 7 8 9 10 11 12
QCM QCM SCM SCM SCM ECM (TCM) PYT SCM ET ET
/QCM /ET
Fret 13 14 15 16 17
QCM PYT PYT QCM (PYT)
/ET
The cittern fret spacings, scaled for a fret 12 at 300mm, can be compared with those shown
above for SCM and QCM:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12 12/13 14/15 16/17
39 25 | 33 25 | 30 26 | 21 | 23 21 | 21 19 | 18 || 19 14 | 14 12 | 14||
| | || || | | || | | ||
The pattern of spacings, with a very large 0/1 is: | VL S | L S |L | L S | L S | L S | L ||
The actual sizes are also practically significant, particularly for the smaller spacings:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12 12/13 14/15 16/17
22 14 | 19 14 | 17 15 | 12 | 13 12 | 12 11 | 10 || 11 8 | 8 7 | 8 ||
| | || || | | || | | ||
A cittern has different open strings from the lute, and appropriate enharmonic choices are
needed. From both the music and the open string pitches the best start appears to be very similar
to the lute G string that had
G Ab A Bb B C Db D Eb E F F# G
The D string will therefore acquire a C instead of D , an A string a further G# instead of Ab, the
# b

E string a further D# instead of Eb, and a B string a further A# instead of Bb. It is seen that
changes to the enharmonics occur on fret 11. A possible difference from a lute could be a Gb on
the G string, so that the D string acquires F#, then the C# appears on the A string, etc. The Db on
fret 6 transforms to Bb and Ab, on courses 1 and 2, and this may be a firmer choice than on a
lute. An Ab on course 2 would meet a G# for fret 4 on the top E course, but this fret is absent on
some early citterns. These differences from a lute result from the position and musical use of the
strings and, for the French tuning, from a smaller number of successive fourths between open
strings.
An average or effective temperament for this English cittern would be about SCM, but there are
some striking deviations. An initial impression may be a wide scatter reaching outside PYT and
TCM. This mainly involves fret 7, and fret 17 which is the octave of 5. Fret 5 itself is almost
exactly SCM, and remarkably like Rose’s. Fret 17 looks very flat on the diagram, at 147.3mm
from the nut, scaled to the lower octave for a 300mm fret 12, as compared with 150.8mm for a
simple PYT. Scaled positions are useful for comparing the very different string lengths. The
discrepancy for the actual string length of around 350mm is about 2mm, but this top fret is not
too important. The crucial fret 7 is also flat at a scaled 198.2mm compared with 198.4 for TCM
and 199.2 for SCM. However, the required actual sharpening for an exact SCM is only about
0.6mm. This is close to the limits of geometry, marking out and making. It can be recalled that
the measured positions only had ½mm accuracy. As with Rose’s frets, any analysis of frets 5
and 7 should involve workmanship and tuning, rather than provide a meaningful guide to
temperament. Similarly to the orpharion, these crucial frets are not too badly tuned for SCM, or
any temperament. The distance of fret 12 from the nut is 171mm, which is significantly less than
half the string length of 175mm. Some uncertainty in the position for the bridge might result
from replacement, its adjustability, curating etc, as much as any stretch-sharpening. It would not
be sensible to sacrifice a reasonable analysis of all these frets in order to retain some nominal
position for the bridge. The other frets can now be analysed in detail.
Fret 1 is almost QCM, which is considerably greater tempering than SCM, or Rose’s ECM. An
octave higher, fret 13 has exactly the same QCM. Ref 1 found that a method similar to Gerle’s
construction could use a frequency ratio of 15/14 for QCM, and 16/15 between QCM and SCM.
Alternatively, since I concluded that no other suitable fractions existed, it was suggested that
Gerle’s original 33/31 for fret 1 could have been used as a general initial position for other
temperings. Further conclusions are given after examining the other lower crucial frets.
Fret 3 is exactly SCM, while fret 15 is between PYT and ET. Gerle’s 99/83 for SCM was
cleverly related to his fret 1 in SCM. Ref 1 found that similar constructions for any intermediate
fret 3 between PYT and TCM were difficult. However, using a construction unrelated to fret 1, I
found an extremely simple 25/21 in ECM. This could be adjusted slightly and used for the
cittern. The construction of 5x5 divisions would be similar to 3x5 divisions for 15/14 on fret 1.
Fret 2 is close to QCM, while fret 14 is exactly PYT. The higher fret is exactly like Rose’s. The
unscaled difference between fret 2 and fret 14, transposed an octave lower, is 2mm. The
difference for a 600mm string would be 3.4mm. Accuracies of the maker and the museum
measurer may be compatible with a fret 2 nearer SCM. However, as noted in Ref 1 this fret
varies only slightly over temperaments, so it is not a crucial guide. It was difficult for the
ancients to construct anything better than a perfect 9/8, as used by Gerle and many others.
Adjustment would be needed later, and perhaps the maker, like Rose, forgot this for fret 14.
Fret 4 is between QCM and SCM, while fret 16 is between QCM and TCM. As noted above, the
fret 5 in SCM may have been an adjustment, or error, from a perfect fourth. Rose’s fret 4 was a
midpoint of frets 3 and 5, Gerle’s was in between, and the cittern’s fret 4 is closer to fret 3.
Similarly, fret 2 is closer to fret 1 rather than Rose’s midpoint of frets 1 and 3. New
constructions for an intermediate fret 4 would be difficult, so a possible method for the cittern
would be a perfect major third of 5/4, which again has a 5 part division as for frets 1 and 3.
Dowland later changed Gerle’s fret 3 specifically to obtain this same position, and it would
probably have needed more adjustment than on the cittern.
It is useful here to make some important interim comments on the possible origins of the first
five crucial frets on a cittern. One explanation might be a variation on Gerle’s scheme of 1532.
However, the earliest known cittern is by Plebanus in around 1536 (Ref 6), and the characteristic
pattern of cittern frets, as seen in the present case and many others, was most likely in existence
some time before this date. In Ref 1 it was concluded that Gerle was supplying rules for a new
generally tempered type of tuning, in addition to an exact SCM. It is possible that he was
influenced by citterns, and then in turn his scheme helped cittern design. At the least, both may
have simply existed in a general atmosphere of meantone, but with no definite links. From the
available evidence one cannot know exactly how Gerle or the citterns first developed from
earlier schemes or more flexible procedures. However, it is not helpful to suggest it was all done
by ear and eye, because this necessarily involves specific ideas of tuning. A modern keyboard
tuner, or a lute player moving frets around, is relying on learnt rules and experience, if not a
precise theory and understanding.
The present cittern has an extreme tempering of these lower frets, but other citterns below are
closer to SCM. The analysis of all the instruments suggests an alternative simpler initial
construction for a tempered tuning. For the first seven frets the frequency ratios are close to:
Fret 1 2 3 4 5 6 7
15/14 9/8 25/21 5/4 4/3 24/17 3/2
QCM (PYT) ECM QCM (PYT) ET (PYT)
This is just an illustration of feasible ancient constructions using more general principles of
tuning. Using such a method, the present cittern would actually need less adjustment than with
Gerle’s scheme. The constructions are simple multiples of 2, 3 and 5 parts, and a midpoint for
fret 6. A slightly flatter fret 1 could use the ratio 16/15. A fret 3 extension from fret 1 could be
135/113 from 15/14, or 128/107 from 16/15, but these are highly unlikely. The difficult fret 3
could be set up explicitly as a midpoint of 32/27 in PYT and 6/5 in TCM, but I am not aware of
any writing on such three part constructions, unless one counts Dowland’s change to Gerle.
Resuming analysis of higher frets, the upper octave can be seen as a reasonable match to the
lower frets. Fret 13 is exactly as fret 1, fret 15 is not bad, fret 16 better, but the slowly varying
fret 14 is PYT like Rose’s, and fret 17 is not well tuned. All this is consistent with the initial
reference of fret 12 for the octave, which is important for the analysis, and particularly the
higher frets of the first octave. Otherwise the whole tuning could be in question, possibly
inviting some complicated transformations for frets 5, 7, 12, 17. The remaining frets 6 to 11 can
now be examined.
Fret 6 between ET and SCM is near the ancient midpoint of 5 and 7, as seen in the lute schemes
and Rose’s frets.
Fret 8 is between PYT and ET, a long way from a perfect fifth on fret 1, and Rose’s SCM.
However, the cittern’s fret 8 is close to Gerle’s suggestion of equal spacings for 6/7 and 7/8.
Fret 9 is exactly SCM and some way from Rose’s ET. It is also a perfect fifth from the fret 2 in
QCM. This is another remarkable instance of a slowly varying fret, most likely constructed in
PYT, but given an intermediate tempering, as seen with frets 2, 3, 4, 5.
Fret 10 is close to ET, some distance from a fifth on fret 3. It is not far from Rose’s PYT, which
did correspond with his fret 3. The ancient schemes had considerable scatter for this relatively
simple fret, which might be expected to vary like fret 2.
Fret 11 is close to ET, and not far from a fifth on fret 4. A fifth on Gerle’s fret 4 would be
similar. Rose’s fret 11 was slightly flatter and more tempered. All these intermediate positions
for fret 11 would serve for alternative enharmonics, as with fret 6.
The appearance of these upper frets in the list of spacings is smoother than Rose’s peculiar large
and small pair of spacings, 10/11 and 9/10. Also, the cittern’s frets have different tempering than
Dowland’s later fifths from lower frets. Rose’s frets are quite close to Dowland, apart from the
latter’s flat fret 11, which may be significant.
As noted above, these higher frets are less crucial than the first seven. It was found in Ref 1 that
the ancients had theoretical difficulty in constructing these frets. Several factors, such as the
simple fifths, smaller spacings for higher frets and the small string length, mean that tiny
deviations in marking and sawing would produce wide variations in the precise pattern of frets
and degree of temper. The effect is similar to frets 2, 5, 7. As one examines several real
instruments, these upper frets appear somewhat sketchy or random, but also generally quite
even. The overall result of analysis on the diagram is a range of tempering between PYT and
SCM. The average is near ET, perhaps more a result of marking and sawing than a desired
tuning. A practical factor favouring this tendency to ET is that a true SCM, and especially QCM,
would require some very small spacings, as seen in the above fret patterns. This would be a
serious physical limitation, and worse than a simple inconvenience of small spaces next to large
ones noted in Ref 2 for lutes. The partial recovery of SCM for the higher octave does not greatly
change this matter. In summary, the characteristic pattern of this cittern is ‘large and small’ with
a tempering between QCM and SCM, followed by a slightly random ET. The carved head seems
like a sphinx staring down its banded chest, challenging one to solve the riddle, but this is
trickier than ‘what has two, three and four legs’.
It is obviously important to know whether this formally inconsistent fretting sounded out of
tune. For now, one possibility is that the lower frets were mainly intended for closely spaced
well tuned chords with good thirds, while the higher frets were for melody and rapid divisions
on the first two courses. There is more on tuning matters below, after assessing three further
citterns, then the effects of inelastic wire, and natural or just tuning. The following treatment of
Italian and Dutch citterns can be shorter. The overall patterns of fretting positions are similar to
the English specimen, but their origins and string lengths are less unusual.
Campi’s cittern
This cittern is in the RCM museum, and has a string length of 430mm. The frets are ‘diatonic’
and extend to fret 19, with only positions 4 and 18 left out. The fret positions were measured by
Ian Harwood and quoted to 0.1mm in Ref 14. These can listed as:
Nut 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Bridge
mm 25.8 45.9 69.2 - 106.8 127 141.5 159 172 188.8 202 214.5 228 237 247.5 257 267.5 - 284 (430)
All 17 frets are plotted on a diagram of temperaments, and the second octave is presented as for
the English cittern (Ec). Perfect fifth constructions for frets 8 to 11 are not included, but the
square in brackets is a fifth below fret 11. For later discussion, the solid dots represent a shift of
the nut 1.5mm away from the frets. There is much similarity to the fret positions on Ec, and a
few significant differences.
As above, it is helpful to show the actual fret positions that would be required for an exact SCM:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
25.8 45.9 69 (87.1) 107.8 127.2 142.3 159.5 173 188.5 200.6 214.3 227.2 237.2 248.8 257.8 268.2 (277.9) 285.4
mm
This shows that the main differences are sharper frets 5, 7, 9, 16, 17 in SCM, by about 1mm.
There is a greater sharpness of 1.4mm for frets 15 and 19. Fret 11 is flatter by 1.4mm, and fret
13 by about 1mm. Fret 12 is slightly less than half the quoted string length by 0.5mm, which
appears satisfactory. Each fret can be compared with Ec in more detail.
Fret 1 is exactly SCM. This tempering, or a slightly greater degree, may be generally considered
to be more common than the extreme QCM position for Ec. However, the higher octave fret 13
is nearer QCM.
Fret 2 is exactly SCM, and fret 14 is nearer QCM. The systems for Ec were QCM and PYT. The
notable feature of both citterns and the orpharion is that fret 2, which would have been difficult
to set by theory, appears to have been adjusted to a tempered position from a simple PYT. In
Campi’s case fret 14 is also tempered, unlike Rose’s and Ec.
Fret 3 is exactly SCM, but fret 15 is some way beyond PYT. As for Ec, the crucial and difficult
fret 3 has been adjusted to the SCM position of Gerle. Fret 15 is even flatter than Rose’s and Ec.
Rose, and Dowland 30 years later, also had fret 3 around PYT. All this indicates that PYT, with
a frequency ratio of 32/27, was indeed a common ancient construction as suggested in Ref 1,
and often left without adjustment. A further simple 6/5 for TCM could have provided the
average at SCM, as explained above.
There is no fret 4, but fret 16 is slightly closer to QCM than for Ec. Similar implications hold as
for Ec, but less strongly for this less crucial higher fret.
Fret 5 is slightly beyond PYT, at a scaled position of 149.5mm. This is not so flat as the fret 17
scaled position of 147.3mm for Ec. An actual shift of 1.0mm would be needed to place fret 5 in
SCM, or 0.5mm between PYT and ET, which are understandable working errors. The present
fret 17 is flatter at an equivalent 149.0mm. The previous discussion of frets 5 and 7 for Ec and
Rose also applies to Campi’s cittern.
Fret 7, in contrast, is slightly beyond TCM at a scaled position of 198.2mm, which is exactly the
same as Ec. The high fret 19 looks very flat at a scaled 195mm, which is 4mm flatter than for
SCM. This measure is relative to the lower octave and a longer string, and is therefore much
larger than the actual 1.4mm seen from the above lists. In simple theory, fret 19 should have half
the sounding length of fret 7, and this could have been a main method of setting up cittern frets.
Fret 6 is SCM, which differs from the midpoint of frets 5 and 7 seen for previous instruments
and schemes. The position is 3mm closer to fret 7 than the midpoint, and not changed greatly by
the above deviations for frets 5 and 7. This interesting difference may have been a deliberate
choice, because the octave and other higher frets are consistent. On more detailed re-
examination there is also a tendency to SCM in Ec, but not in Rose.
Fret 8 is between ET and SCM. This is flatter than a perfect fifth from fret 1, but sharper than
the equal 6/7 and 7/8 of Gerle. Both Rose and Ec had equal spacings, but Ec had a flatter fret 8
in PYT.
Fret 9 is close to QCM, but not a perfect fifth from fret 2. It is similar to Ec, but flatter than
Rose. This is a further case of providing a tempered fret instead of the basic PYT.
Fret 10 is between SCM and QCM and a good fifth from fret 3. It is sharper than Rose in PYT
and Ec near ET.
Fret 11 is near ET, and close to both Ec and Rose. As noted above, the intermediate position and
tempering would serve both enharmonics. A fifth below fret 11 could have produced a
reasonable fret 4 near ET. A fourth below fret 16 in QCM, or a fifth above an equivalent fret 4,
would lead to a similarly flatter fret 11 in QCM.
The fret measurements were quoted in an early study (Ref 14) that gave deviations in cents from
ET. A comparison with the short table on p15 of Ref 2 translates these deviations into an almost
exact SCM tempering. However, frets 5, 7, 12, 19 were adjusted to be more perfect, and an error
of about 1.3mm in fret 10 and most higher frets was suggested to result from an ancient mistake
with a strip. There is no special sign of this in the present diagram, where a similar inaccuracy
appears for several frets between 5 and 19. More importantly, there may be other geometric
uncertainties in making, and complications from inelastic wires, so the present analysis has
preferred to use only fret 12 as an initial reference.
Brussels - KM 1542; Ashmolean - Virchi ?; and Gasparo da Salo
The Brussels cittern is thought to be Dutch, and loosely dated as 16 or 17th century (Ref 11). The
string length is 433mm, similar to Campi’s. The top three courses appear to be double and the
fourth triple. The fretting is diatonic and extends to fret 19, with frets 4 and 18 missed out. Frets
6, 8, 13 appear to serve only the top two courses, and fret 11 the top three. The only data found
were photos (Ref 5), and measurements of frets may be reasonable to only about plus or minus
1mm. The fret positions are potted on a diagram, using circles with crosses and squares as
previously.
A cittern in the Ashmolean Museum catalogued as anonymous and dated 1570 may have been
made by Virchi (Ref 11). The string length is 462mm, a little longer than the others, and there
are six double courses. The fretting is fully chromatic with 19 frets under all courses. Fret
positions taken from a good photo (Ref 15) are plotted with the Dutch data and have similar
accuracy. Frets 1 to 12 now use plain crosses and the higher frets use plain circles. These two
citterns are included mainly to show a more representative sample. The overall fretting patterns
and tuning are very similar to Campi’s, and can be discussed briefly together.
The Ashmolean has one other cittern dated 1560-70 and made by Gasparo da Salo. It has fully
chromatic frets, and the eleven pegs could probably be assigned to five courses of length
453mm This instrument is mentioned because the frets have an obvious, perhaps perfect, ET
pattern. The fingerboard is thought to be a replacement, and such citterns were produced into the
1700s. ET tuning is not favoured by modern makers, and possible reasons will be examined
below. A late 18th century cittern can look very modern with six courses, twelve simple frets in
uniform ET, and sometimes a small keyboard to pluck the strings.
Fret 1, for the Dutch cittern and for Virchi’s, is close to SCM, but fret 13 is nearer ET, similar to
the tendency noted above for higher frets.
Fret 2 is QCM, again showing a tempered adjustment to PYT, and fret 14 is similar.
Fret 3 for Virchi is close to a tempered SCM, but the Dutch position appears even sharper than
TCM. Both have fret 15 at TCM. The spacing 2/3 does not look as large as on Ec, but 2/1 is not
so small since fret 1 is flatter and fret 2 sharper.
Frets 4 and 16 for Virchi are near ET, unlike the other citterns, but subject to inaccuracy. The
Dutch cittern lacks a fret 4, but fret 16 is TCM, not far from QCM positions for Campi and Ec.
Fret 5 for both citterns is close to a simple PYT, but fret 17 is flatter by about 0.7mm.
Fret 7 is PYT in the Dutch case, and about 0.7mm sharper for Virchi. Both have fret 19 too flat
by about 1.5mm. These deviations for frets 5, 7, 17, 19 are similar to the other citterns.
Fret 6 is near SCM for Virchi, and is closer to fret 7 than the midpoint with fret 5, as seen in
Campi. The Dutch fret 6 is slightly closer to the intermediate ET. Fret 18 for Virchi is nearer to
PYT, but missing in the Dutch case.
The upper frets 8 to 11 are more affected by the uncertainty of measurement. Visually there are
similarities in the patterns for all four citterns. For example, there are relatively larger spacings
for 7/8 and 9/10, but this in turn depends on frets 7 and 10.
Fret 8 may be between SCM and QCM, and more tempered than Campi and Ec. Fret 9 in ET is
less tempered. Fret 10 is like Campi and more tempered than Ec in PYT. Fret 11 is close to the
average ET of the other citterns.
In summary, these two citterns show the same distinct pattern of SCM for the lower four frets
with a tendency to QCM, and some tempered adjustment of difficult constructions for frets 2
and 4. Frets 5 and 7 have significant but practically understandable deviations, and fret 6 is
biased to SCM. The upper frets are also scattered around SCM, but with some random tendency
to ET, particularly for fret 11.
The missing or partial frets 4, 6, 8, 11, 13, 18 on the Dutch cittern would be similar to the
accidentals not used in a mode based on D. The missing frets for a mode based on a D string
would be 1, 4, 6, 8, 11, 13, 16, 18. A notable difference is the presence of the important fret 1,
and only frets 4 and 18 are completely missed on citterns. Other open strings could be related to
different modes, such as a G string missing frets 1, 3, 6, 8, 11; and an A string would miss frets
1, 4, 6, 9, 11. These differ from the D string by a single fret, but not fret 1. An E string could
miss frets 2, 4, 6, 9, 11; and a B string frets 2, 4, 7, 9, 11. For completeness a C string misses
frets 1, 3, 6, 8, 10; and an F string frets 1, 3, 5, 6, 8, 10. These considerations indicate that fret 4
is missed simply because it has sharps of limited use. Much tablature music does not use fret 4.
There are still many major thirds between strings even if this interval is removed from single
strings. On a lute the sharper notes on fret 4 have many uses, and the difference for the cittern
depends on the positions of the open string tunings, as indicated above. Frets 6, 8, 11 are more
like those on the lute. In view of these details, the term ‘partially diatonic’ seems more accurate,
but similar to ‘partially vegetarian’.
Regarding Dutch citterns, a famous depiction appears in Vermeer’s ‘The love letter’ of 1669. A
woman and her maid are glimpsed in a distant room at the end of a corridor. The painted image
of the cittern is small but frets can be discerned, and it is thought that Vermeer used optical aids.
The first four or five frets are obscured by a hand. This is very common in lute paintings, and
disturbs an artist’s grasp of geometry. Above the hand are 13 frets. Very broadly, there are five
almost equal pairs of spacings, the first pair ‘small’, the second pair ‘large’ and the next three
pairs successively smaller. The next space is again smaller followed by a larger gap, maybe for a
missing fret 18, before the top fret. The pattern appears similar to ECM, but not ET. However,
in addition to the first small pair there is a slight alternation of spacing in the other pairs, which
conveys the appearance of real cittern frets. In contrast, Vermeer’s ‘Woman with a lute’ and
‘The guitar player’ from 1663 and 1672 both have definite ET patterns. However, the former has
a small spacing next to the nut, and it was a common misperception that the first spacing
included the nut. The ET frets may strengthen credibility in his cittern frets, and for a late 1669.

WIRE ELASTICITY and STRETCH - SHARPENING

Modern makers of wire-strung fretted instruments are very familiar with large rises of pitch
when a string is pushed down to a fret. The effect is so large that special care is needed in design
and the whole procedure of making. This includes the fingerboard, setting up the fixed frets, and
the action height of the strings. Similar effects occur with a lute, but are not so extreme or
sensitive. The difference depends entirely on the much lower elasticity of metal wire compared
with gut or modern synthetics. As early as Ref 2, I noted that the effect would increase as the
square of the action height and could be calculated in detail. This was done for Refs 12 and 16
while examining how stretch-sharpening might be implicated in angled bridges on lutes. It was
demonstrated that such elastic considerations were physically and practically impossible, but
that string taper was the decisive mechanism. However, some sample calculations were given of
stretch-sharpening for different heights of frets and strings, but cittern makers may have been
disappointed that the complete theory was not included. Before giving the details here, an initial
quantitative comparison can be made with a lute. This will show quickly how the theory has
very practical uses.
When a string is pressed down by a typical action height a it is stretched by a length x
proportional to a2. This may not be immediately obvious but follows from geometry, and is
explained below. For a lute we are familiar with the need to have an action height of say about
3mm near fret 9. The increase of pitch from this stretching depends on the stress being increased
by an amount proportional to Ea2 where E is the tensile modulus. If we have a lute and a cittern
with the same string length and at the same initial pitch, then for the cittern we can find the
action height that will produce the same pitch rise as the lute. As seen below, this is given by
(am /ag)2 = (Eg /Em)(m /g), where the subscripts indicate the metal strings of the cittern, and
gut or a substitute for the lute. It is important to include the density  in the analysis. This
relation is general and does not need to specify any particular string length, diameter, or
frequency. Typical values for Eg, Em, g, m are 3GPa, 100GPa, 1.3gm/cc, 8gm/cc. This gives
(am /ag)2 = (3/100)(8/1.3) = 0.185, or am = 0.43ag. The wire strings are predicted to need an
action height of about 1.3mm, which is less than half that for lute strings. For stiffer metals such
as steel the modulus may be 200GPa, and the height is reduced to 0.9mm. The theory
corresponds well with the tolerances and care needed in making a cittern, and also gives lute
makers an idea of the extra problems. Substantial changes of pitch can be anticipated and are
calculated below.
As shown in Ref 12, there are contributions to the total stretching of a string from different parts
along its length, and all of these can be calculated. There is a similar geometry for both lutes and
citterns. The main difference is that metal frets are fitted level with the fingerboard surface, and
then wood is scalloped out to leave the frets as cusps with a gradual sloping to a lower level. On
a lute the later fitting of gut frets produces a more abrupt step. A schematic drawing shows the
geometry, and for clarity the scales are grossly distorted. The actual stretch, of order 0.1mm,
could not be found from a drawing, and calculation is necessary. The required dimensions are:
n - the height of the nut above the fingerboard, or the base of a scallop; m1, m2 etc - the height of
each successive fret. The symbol m is used to match Ref 12, and m0 could be used for the nut,
but n is more distinctive. On a cittern the fingerboard is raised above the belly, so the string
height a at the bridge is taken relative to a projection of this higher level. The distance between
the nut and bridge is L, and the slightly larger initial length of the tuned string between the tops
of the nut and bridge is L0. The distance of a fret from the nut is ln , and in analysing the nth fret,
the lower fret n-1 will also be important. The calculations involve repeated applications of
Pythagoras’s theorem.
First the length of the open unfretted string can be found from L0 =  (L2 + (a – n)2 ). Since L is
much larger than a and n, this can be simplified to L0 = L + (a – n)2 /2L. This is a small
distinction from the nominal L, and takes account of different values for a and n. Relations of
this form can be applied to the other hypotenuses or diagonals formed by the fretted string.
When the string is lightly pressed to the top of fret n, the new length of string Lp is
(L - ln ) + (a – mn)2 /2(L - ln ) + ln + (n – mn )2 /2ln
--------------1st--------------- -------2nd--------
The 1st part is the diagonal between fret and bridge, and the 2nd part is between fret and nut. The
pitch increase here is due to the action height only. Even on a lute one can detect a slight effect,
although this light pressing is not the usual way of fingering a lute.
When the finger is pressed down between fret n and the lower fret (n-1) the string is further
stretched to a length Ls given by
(L - ln ) + (a – mn )2 /2(L - ln ) + ln -1 + (n – mn-1)2 /2ln-1 + (ln - ln -1) + (mn2 + mn-12) /(ln - ln -1)
--------------1st--------------- ----------2nd----------- ------------------3rd-------------------
The 1 part is unchanged, but the 2 is now between fret (n-1) and the nut, and the 3rd
st nd

expression arises in the spacing between the two frets. On a lute the increased effect from the
stronger more normal pressing is noticeable. Ref 12 presented this 3rd component as an
illustration of the theory. Ref 1 had anticipated this could be significant, especially with a small
spacing between frets as in SCM. The simplest geometry has been used to estimate this 3rd part.
This is just two lines sloping down to the centre of the spacing. The finger will flatten the
contact, producing even more stretch, and these details are treated later. This stretching between
frets can be much larger than in light pressing, as a result of the small denominator (ln - ln -1).
The stretch x caused by light pressing, Lp - L0, is given by
(a – mn)2 /2(L - ln ) + (n – mn)2 /2ln - (a – n)2 /2L
The full stretch caused by hard pressing, Ls - L0, is given by
(a – mn )2 /2(L - ln ) + (n – mn-1)2 /2ln-1 + (mn2 + mn-12) /(ln - ln -1) - (a – n)2 /2L
These relations together with the following elasticity could be called the cittern equations, since
the effects will be much larger than for a lute, and they are worth some detailed discussion.
The expressions can be applied to several typical string lengths, fret positions, fret heights and
action heights. From a short compilation (Ref 17) a string height above fret 1 may be around
0.5mm, and above fret 17 about 1.5mm. This produces the following heights above the fret tips
of about 0.5mm near to the nut, 1.3mm at fret 12, 1.5mm at fret 17, and 2.1mm at the bridge.
One can recall the above theoretical estimate of a 1.3mm string height at fret 9 being equivalent
to a height of 3mm on a lute.
Also in Ref 17, Peter Forrester reported that his scallops between frets were flat-bottomed with
depths of 0.6, 0.75 and 0.8mm for string lengths of 43, 61cm and a bandora.
Therefore, with the 0.6mm depth the nut height n is 1.1mm, all the fret heights mn are 0.6mm,
and the effective bridge height is 2.7mm. This can be called data set 1.
For the 0.8mm depth all the heights are just 0.2mm greater at 1.3, 0.8, 2.9mm.
For a preliminary example I had chosen round figures of 1.0, 0.5, 2.0mm. These are lower, but it
will be seen later that all the above values of n are high, due to the initial estimate of a 0.5mm
string height near the nut. A host of possibilities could be calculated, but it will be better to
restrict the exercise fairly close to practice, and the main dependencies will become clear.
For data set 1 the values for L, n, a, m are 430, 1.1, 2.7, 0.6mm, and fret positions are taken for a
theoretical SCM. It is instructive to list all the separate components of stretch, rather than a
simple final amount. The results are given in units of 0.01mm, and a small amount, 0.003mm
for (L0 – L), needs to be subtracted from each total stretch.
Fret no. Fret to bridge Fret to nut Between frets Total ET
(0.01mm) hard press light press hard light between
open string 0 0 0 0 0 0 0
1 0.546 (6.04) (0.481) 6.0 6.6 1.03 6.9
2 0.574 0.483 (0.271) 3.6 4.6 0.84 3.2
3 0.611 0.271 (0.181) 3.1 4.0 0.79 3.3
4 0.643 0.181 (0.143) 4.0 4.8 0.78 3.6
5 0.685 0.143 (0.116) 3.4 4.2 0.80 3.7
7 0.768 0.096 (0.087) 4.8 5.7 0.77 4.3
12 1.025 0.062 (0.058) 5.1 6.2 1.09 5.5
19 1.542 0.045 (0.044) 9.6 11.2 1.58 8.5
The largest component comes from pressing down between the frets, and increases from
0.03mm for fret 3 up to about 0.1mm for fret 19. The effect for fret 1 is large due to the rather
high nut. A nut can be viewed as a fret 0, and an extra fret is often placed in front of the nut.
After fret 1 the alternating values are due to the large and small spacings in SCM, and the
smallest stretching arises for the large spacing 2/3. Calculations of stretch between frets were
also made for the different spacings of ET frets. The stretch is greater for fret 1, less for fret 2,
slightly greater for fret 3 and then increases steadily up to large values for frets 12 and 19.
Calculations show that missing out a fret such as 4 would greatly reduce, to less than about a
half, the stretch associated with the higher fret 5, as anticipated above.
Next we need to find the effect of the extra stretch in increasing the string tension. This can be
deduced as a general measure without reference to a specific wire gauge, tension T or pitch f.
The frequency f is given by
f = ( (/) ) / 2l
where  is the stress T/A in the string, A is the cross section area ¼d2 for a string of diameter
d, and l is the sounding length.
Before our main concern, the relation between f, , , l can provide a neat explanation of the
unexpectedly fragile top metal strings. For gut and metal of equal length to break at the same
pitch would require equal values of (g /g) and (m /m), where the subscripts also denote the
breaking stress. We know that m is close to 6g , so that m would need to be 6g. This is only
true for very strong metals such as piano wire, but for the softer metals m may be no more than
about 2g. Synthetic alternatives can have even higher strengths than natural gut. This may be
clearer than comparing detailed values that can vary widely between materials. A more practical
working limit may be the stress for initial or more significant yield, for which the explanation is
also applicable. In practice, the limit for the standard metals is so severe that a shorter thin string
can be very useful, and a 10% shortening will increase the pitch by a whole tone.
Now we can find the increase in stress when the open string is stretched by pressing down to a
fret. The simplest realistic assumption is that the increased tension is distributed along the entire
string and not confined to one section. (Other distributions would be difficult to predict, but
some interesting detailed effects will be mentioned later. It is unlikely that the transient pressing
would raise the tension in the further lengths between bridge and tail, or between nut and peg.
Over time, repeated playing would contribute to a gradual loss of tuning.) The first step is to
calculate the increase in stress that a small stretch of x would produce on an open string. Then
for each fret the calculated stretch will give a raised stress acting with the relevant sounding
length.
For an open unfretted string
f = ( (/) ) /2L0
The change in pitch f when the stress is increased by the small amount of additional stretch x is
given by f /f = ½ /. The initial stress is given by o = 4 f 2 L02, and the increase in stress is
 = Ex /L0. The relative increase in pitch is therefore
f /f = xE /(8 f 2 L03)
An effective initial stretch needed to bring a slack open string up to pitch could be defined as
xo = 4 f 2 L03/E.
Depending on linearity this would be more or less close to the actual amount of winding on a
peg. The relative rise of pitch can therefore be written simply as half the relative increase of
stretch
f /f = x/2x0.
For the fretted notes on a given string the actual pitches can be given formally by the relation
f = (1 + f /f )((/))/ 2l
where l is the length between bridge and fret, (L - ln ) + (a – mn)2 /2(L - ln ),  is the initial
nominal stress, and the fractional frequency rise is calculated as above for an open string. For
cases where one wants to assess the effect of elastic stretching on a nominal pitch, either on a
real lute or a theoretical scheme, then the factor (1 + f /f ) is sufficient. This saves making a
new calculation for each fret and string. If a more specific relation is desired, then the usual
f = ( (T/) ) /d l
could be used for the basic uncorrected pitches, but this is unnecessary.
Now the theory can be applied to real cases. For the 3rd course g string on a cittern, the values
for , f, l, L0, E can be taken as 8gm/cc, 196Hz, 43cm, 43cm, 100GPa. The predicted relative
rise of pitch is f /f = 512x, and the effective initial stretch is 1.0mm (by chance). This can now
be applied to the fretted notes in SCM.
For hard pressing, the lowest value of x is 0.04mm on fret 3, or 4.0 x 10-5m, so the relative rise
of pitch is f /f = 0.020. The largest value of x is 0.11mm on fret 19 giving f /f as 0.06, which
is equivalent to a frequency ratio of 1.060. A semitone in ET is 1.0595, and this is 100cents. A
syntonic comma is 81/80 = 1.0125, or 22 cents. The predicted pitch rises are therefore large, and
of order a third to a whole semitone. In view of this enormous effect, I needed to test a brass
string, and found the same low initial stretch of about 1mm. This also involved various loads
and wire gauges. Many data tables confirm a typical modulus of 100GPa, largely independent of
yield stress and breaking point, both of which have more variation. (After working on theories
of metals under extreme conditions it seems odd to be so concerned with this mild regime.)
In contrast, the extra stretch x for light pressing to a fret is much less, and varies from 0.007mm
at fret 1, down to 0.005, then up to 0.013mm at fret 19. These stretches produce much lower
pitch rises of about 6, 4, and 10cents.
For each different fretted note on the same string, the appropriate value of x gives the sharpened
pitch. The nominal sounding length for a fretted note would be (L - ln ). The precise extra length
is (a – mn)2 /2(L - ln ) and in the present case this has a value of 0.005mm for fret 1 rising to
0.015mm at fret 19. Similarly, the open sounding length L0 was only 0.003mm greater than L.
These are completely insignificant as corrections to the nominal sounding lengths, but it was
sensible to retain them until the dominant elastic effects had been evaluated.
The consequences of this theory can be examined for the low 3rd course g string, and then the
higher strings and octaves. For hard pressing at fret 1 the pitch rise is 1.032 for SCM, which is
over half a semitone, and 1.034 for ET or about 3 cents higher. For fret 2 the rises are 1.022 and
1.015. These rises of a third and a quarter of a semitone are smaller than for fret 1 but differ
more from each other. Pitch rises can be translated into equivalent changes in fret position. For a
typical fret spacing of 30mm the changes are of order 10mm, which is enormous. Even the
relative deviations for different tempering are large, such that fret 2 in SCM is sharpened 3mm
more than an ET fret. From the above table, these effects decrease in the middle of the first
octave but then double towards fret 19. This needs serious investigation, since real citterns could
not operate under such conditions. It will be seen that several factors can soften the above
predictions, but still leave an instrument highly sensitive to its stiff strings. For some practical
feel during this study, I have produced an ersatz orpharion by modifying a small guitar.
For the low g string on a 43cm cittern, makers would use brass with an effective diameter of
0.018inch, 18thou or 0.46mm. The above equation leads to a tension of 3.8kg. The density of
8gm/cc has been taken as representative of both brass and iron, but a high copper density of
9gm/cc would require a tension of 4.3kg. Both these tensions are rather high but could be
reduced to a more reasonable 3kg by using a thinner 16thou gauge. Cittern makers know that
solid brass wires cannot be used above about 16thou, since inharmonicity produces an
unacceptable dead sound. This occurs quite abruptly at a certain diameter, and was well known
to makers of harps, harpsichords and pianos, before the more recent revival of interest in gut.
The effect is related to the present concerns about elasticity and was discussed in Ref 18, but
will not be treated further here. A pair of twisted brass wires with an effective diameter of
18thou would be used for the g string. However, even if the tensile modulus were halved, the
above stretching effect would only be reduced in the same proportion as x/xo or xE. A softer
brass might give only a slight increase in stretch. For a single wire with an acceptable gauge of
16 or 14thou the original prediction is essentially unchanged.
Other factors that could reduce the high sensitivity to stretch are lower string heights. All the
components of stretch increase as the square of a height so that lowering the nut height n could
ameliorate a strong effect from fret 1. With a value of 0.85mm for n, from a smaller string
height of 0.25mm, the large 0.06mm stretch is reduced to 0.042mm, similar to the other frets.
For the lowest possible n, close to the fret height m, the stretch would be 0.028mm. Reducing
the effective action height a towards the bridge would not be much help, because the largest
effect arises from pressing between frets. The height of 0.6mm for the mn is authoritative, and
the only conceivable explanation would be a type of fingering that does not press down fully
behind a fret, but tends to the condition of light pressing for this sensitive string. This would be
possible on the scallop sloping gently down from the fret tip. Later citterns without scallops and
the orpharions had a thin ‘fret wire’ placed here for the same purpose. By pressing nearer to the
fret the effective height m could be reduced smoothly to any fraction of the full 0.6mm, and
produce greatly reduced stretches, merging into the condition of light pressing. With these
smaller effects it would also become useful to reduce the action height towards the bridge.
There could be many variations on the simple geometry of hard pressing used above. Flattening
the string between frets would greatly increase stretch, particularly as the ends of a ‘square well’
are approached. Here the stretch would tend rapidly to 2m, which is a very large 1.2mm, rather
than 2m2/(ln - ln -1) of order 0.04mm. Milder versions of this could be useful on the less sensitive
higher strings, and for expression on a lute. The scallops or fret wire would help control and
avoid disasters with this type of effect. A surprising effect occurs when a finger is slid or rolled
towards the upper fret. The pitch often remains unchanged whereas one might expect
sharpening. Conversely, moving the finger towards the lower fret does sharpen the note. An
explanation is that both movements cause stretch, but the former can also lock the string on a
fret, whereas the latter helps to release the string over the fret and so raises tension in the
sounding length. Another related feature may be the wedges for fitting the frets. In old citterns
these are generally packed behind a fret. Presumably these could be adjusted to flatten a fret, but
any sharpening would appear to be left to finger pressing.
After these complications it is easier to treat the other less sensitive strings. For the octave g
string xo is 4mm, since it varies as the square of the frequency which has doubled. The string
might be iron for tonal and strength reasons, and would have a gauge of about 9thou, in
proportion to 1/f. The lower density also allows a slightly thicker string. The values of increased
stretch x remain exactly as calculated above. Pitch rises for hard pressing at frets 1 and 12 were
of order 50cents, and are now reduced to 12cents. For intermediate frets the rises are now about
8cents, and for the high fret 19 about 25cents. For light pressing the effects are small at about
1½, 1, and 2½ cents. At this stage it can be noted that the low 3rd course would not be used
much higher than fret 7, but it needed to be included above as a reference for other strings.
These three levels of stretch, medium, low and high, can now be applied to the 4th course. The
pitch rises are decreased slightly by a factor of about 0.8, since the ‘a’ string is a tone higher,
and the brass string would have a gauge of about 16thou. The stretching effects of an 8thou
octave would be slightly less than for the g. A pitch of b would produce further decreases of
about 0.8.
The 1st course e strings are usually iron and would have a gauge of about 11thou, from 18x5/8.
The effective initial stretch xo is 2.8mm, from (5/3)2 x1mm The pitch rises are a factor of 0.36
those for the low g string, or 1.44 greater than for the octave g. For hard pressing the three
levels of pitch rise are about 17, 11 and 35cents. For light pressing the rises are about 2, 1½ and
3½cents.
For 2nd course d strings in brass the gauge is about 12thou and the pitch rises are slightly higher
than for the 1st course by a factor of 1.2, or (10/9)2. For the upper two courses the fingering on
higher frets would still need to be light. The small spacings would tend to enforce this, apart
from the need to avoid pitch rises, and it would also be natural for rapid playing.
In summary, these predicted stretch-sharpening effects are relatively large and appear to
necessitate only light pressing on the higher scallop just behind a fret. The amount of sharpening
is greatest for the thicker strings, and for fret 1 followed by frets 12 to 19. A further notable
feature of all these citterns, and also Rose’s orpharion, is that the frets are not slanted towards
the nut to give some compensation for greater stretch-sharpening on the lower strings. This also
has implications for preferred methods on lutes.
After reaching these quantitative theoretical conclusions I found from players that they did
indeed lighten their touch behind a fret, particularly on the low 3rd course. This had become an
instinctive part of technique and they needed some reflection before recalling their precise
actions. In effect, the present theory has deduced from first principles how a cittern works.
While I have some confidence in the elasticity and theoretical geometry, it may be more
appropriate for others to refine the elastic modulus and detailed heights of frets and action to
produce precise comparisons with practice. The theory could assist making, and it may be
possible to make design calculations of sufficient accuracy to fine tune individual frets. As a
general guide, stretch-sharpening effects can be reduced to a quarter by halving string
deflections, and to about a tenth by lowering deflections to a third.
Effects of stretch-sharpening on ancient fret positions and cittern making
The stretching effects have further consequences for the positioning of frets, and hence the
present analysis of the ancient tunings. In addition, they can provide some understanding of the
specialized procedures in making a cittern. A large part of making a fine cittern would involve
small adjustments to frets, scallops and action to enable good tuning. The two aspects can be
taken together since there are many subtle connections. Peter Forrester specializes in making
wire-strung instruments and has several personal approaches. These may seem unusual to
makers who work on a broader range of lutes, and the analysis will engage with his methods.
On the diagrams of fret positions and temperaments, there was considerable centring on an SCM
system, but with a tendency to a more tempered fret 1; some poor tuning on the important frets 5
and 7; and scatter on the higher frets. Initially, it seemed that the elastic effects could produce
large complicated changes, resulting in very different effective tunings. However, two factors
removed this apprehension: the necessity and practical use of light pressing, and also finding
that the ancients had carefully adjusted their frets 2, 3, 4 to a surprisingly consistent tempering.
It was seen in Ref 1 that the ancients had difficulty in giving accurate initial constructions for
these three important frets, so the positions on citterns were most likely found by small
adjustments to their initial constructions. It can be anticipated, and is shown below, that
applying any large changes resulting from stretching to the fret positions generally seen on old
citterns would have been chaotic and disastrous.
The effects of pressing at frets can be seen in the following way. The first key fact is that an
open string is greatly different from any fretted note, as seen from the first line in the above
table. The open string is simply not affected, except for an initial plucking that rapidly settles
down as discussed in Ref 12. In contrast, any fretted note can be sharpened by a significant part
of a semitone. In order to apply a ‘correction’ for this effect, a typical length could be subtracted
from the origin or nut, on the diagram or an actual cittern. Equivalently, the pitch at a fret could
be described by adding a length from the nut. For example, if the stretching gave a pitch rise of
6 cents on fret 2, then the effective position would be further from the nut by about 2mm, for a
spacing of 30mm. This behaviour was considered in the initial attempts with the orpharions, and
is plotted on the diagram of Rose’s fret positions with solid dots. There is an opposite effect on
the effective temperament of alternate frets, such that fret 1 is moved slightly towards QCM,
then fret 2 greatly towards PYT, and so on. The general effect is to distort and scatter a
reasonably consistent system of ECM, but not greatly change the average. This sparse diagram
shows the effect clearly, and the diagram for Campi’s cittern has similar behaviour. Here a
slightly lower shift of 1.5mm was able to position the problematic frets 5 and 7 in reasonably
tuned positions between ET and QCM, but at the expense of disturbing most other frets. The
same effects appear later with Palmer’s frets. An important result of this analysis is that the
sharpness and high tempering of fret 1 is confirmed and potentially increased, rather than
reduced to the average SCM or lower.
Several other systematic transformations were tried, representing different physical and
geometric effects, uncertainties and errors. None made a significant or convincing improvement,
and the actual fret positions are best left unprocessed. However, it is valid and useful to estimate
and discuss the possible effects on each individual fret, but it could be very misleading to
subject fret positions to some ‘best fit routine’. Many decades ago a kindly elderly physicist and
zoologist from Queens’ College once helped me out with ‘statistics is best used as a drunkard
uses a lamp post - for support rather than illumination’.
In Ref 17 Peter Forrester outlined how he set up his tuning. The fingerboard is planed, and often
to a slightly concave longitudinal shape to accommodate the string vibration, which shows the
tight tolerances compared with a lute. Then positions of the nut, bridge and fret 12 are arranged,
using thin 0.25mm wire as a temporary nut. The effects of action height are compensated by
moving the bridge away from the nut. This can be seen in the above theory, since with a very
low nut the amount of stretch increases from fret 1 towards 12, so the correction for stretch-
sharpening also increases. Therefore, the distance from the bridge to fret 12 needs to increase
proportionally more than the bridge to fret 1. A similar smaller effect occurs with a lute, but
instead of moving the bridge the higher frets are shifted slightly more towards the nut.
When the frets have been fitted flush with the fingerboard, and after some scalloping, the ‘nut
wire’ is moved towards the frets by about 2mm until tuned reasonably. A meter is used on the
2nd course, which as seen above has moderate sensitivity to stretch, since d is midway between
g and the octave b. For unison courses the 4th course b would be more central. The movement
of the nut towards the frets is a practical use of the large difference between any fret and the
open string, as seen in the analysis. This point may have puzzled makers with a primary interest
in lutes, where the sharpening effect of stretching between frets is much smaller. On a lute the
action height effect increases for higher frets, as calculated in Refs 2 and 12. For the smaller
effect between frets, any slight increase from a smaller spacing of higher frets is offset by using
thinner fret gut with a smaller height m. On a cittern, a 2mm shift produces a significant
sharpening of an open string that can be balanced by moderate pressing at frets. Finally, the nut
or an extra fret replaces the wire, crucially at the same height, and then the scalloping is
completed. This whole procedure involves a systematic series of increasingly fine adjustments.
Simply setting fret positions with a meter would require some prior testing of the stretching
effects, or extremely light pressing on shallow scallops. It is interesting to see how thoughtful
craftsmanship is often more consistent with a theoretical view, rather than elaborate technical
methods. A simpler example is gluing up any kind of frame, as needed for fine furniture and
basic joinery. For a speedy accurate job one tests simply for equal diagonals, as required by the
geometry of parallelograms, rather than struggles with set-squares while the glue sets.
A final adjustment of the fingerboard is left till an instrument has been strung for some weeks at
full tension. The small tolerances on the action require the whole structure to settle. Regarding
structural stability, a cittern has two bars under the belly, and the new analysis for lutes and
guitars in Ref 18 would apply. For small bodies, the bending of the belly would be less crucial
than a sufficiently sturdy slender neck and joints. For orpharions the seven bars under the belly
would operate as in the analysis for a lute. Extra strength might be provided by the shallow
bodies and their wavier outline. The early gut-strung guitar and vihuela had only two bars, but
the greater ‘lute’ barring on the wire-strung orpharions could provide more stability, which
would be needed for the sensitive action.
Another intriguing aspect of cittern making is the conclusion (Refs 11, 17) that ET frets do not
work well, and that a more tempered system is better. From the present theory it would be
expected that any meantone, including ET, would be necessary and sufficient for a consistent
tuning. Frets in ET were actually used successfully on later citterns, and on some modern
versions, including kits. This indicates that any problem would be a second order detail,
possibly due to elastic effects. On detailed inquiry (Ref 6), a significant problem with ET is
arranging fret 2 to produce: (i) an e on the 2nd course d string to be in unison with the open 1st
course e, and also (ii) an f# on the 1st course that is a good major third to the open d string. To
a good approximation, both should be possible since fret 2 gives an interval of a whole tone in
the respective system of ET or say SCM. However, the unison e on two courses appears in
music very frequently and needs to be accurate, and a major third is purer in SCM than ET. It
may be desirable that closely spaced cittern chords, especially on the most used lower frets and
all four courses, are very well tuned. Sensitivity to pressing would also appear on higher frets,
but for thinner chords. The problem may also arise from the use of only one fret together with an
open string, whereas two fretted notes allow more balancing of touch. In further detail, different
pressings on the two courses can be examined. If some moderate pressing is used to produce an
e on the second course, then a harder pressing would be needed for the first course f#, as
discussed above. In SCM the third is purer and because of the small fret spacing 1/2, the
sharpening effect is stronger and acts quickly. In ET, by contrast, the third is sharp but any
further sharpening effect is weaker, slower and could feel uncertain. This may be a basis of an
explanation, but the precise details of a set-up would be important.
A final instance of adjusting the tuning of a cittern appeared in a lengthy modification of a kit
provided with ET frets (Ref 17). The pre-cut positions were found to produce poor tuning,
especially for frets 2 and 3. In the altered cittern, fret 2 was eventually moved nearer to the nut
by 2mm, and fret 3 by 2.5mm. For ET on a 600mm string, the initial positions and changes can
be deduced as:
Fret 1 2 3
Initial - mm 33½ 31½ 65 30½ 95½
Modified 33½ 29½ 63 30 93
Nut moved 35½ 29½ 65 30 95
ECM 35 30 65 31 96
Spacings between frets are shown by the intermediate numbers. It is seen that the modified frets
have almost equal spacings, and this indicates a move towards greater tempering than ET, as
seen in the section on fret patterns. In the third line 2mm has been added at the nut, to reflect an
effective subtraction that might be needed to account for stretch-sharpening, and the result is
very close to the theoretical ECM in the last line. The changes effectively made the tuning a
more tempered meantone, but further analysis would require fret heights and other details.
It is significant that for this kit the action height at the nut was quoted as 0.5mm. This was the
value deliberately used in the calculations above to show the large amount of stretch sharpening,
particularly for fret 1. A lower value of 0.25mm was better, and it is likely that the initial ET
frets and tuning could have been retained simply by slightly lowering the nut. This would have
been consistent with other modern experience and 18th century citterns. The modification of this
kit is a good demonstration of the accuracies required in a cittern. The fine adjustment needed
for the string height near the nut end is a crucial factor.
A further interesting conclusion from this modification is that an ECM on lower frets appears to
have been compatible with the initial ET on higher frets. A greater contrast between SCM and
even QCM, to a rough ET was found in the analysis of the ancient frets. This gradual change of
tempering higher up the fingerboard is different from a wide scatter on adjacent frets and the
general criterion for a consistent tuning. Similar trends could be tested and used on modern
versions, in addition to the more usual standard SCM.
Nevertheless, there appears to be some variability in the treatment of the nut position in
compensating for the unstretched open string. (During the initial stages of analysis and making
an ersatz orpharion, I even began to wonder whether players carried a small metal strip that they
could insert next to the nut.) In analysing the orpharion and citterns, the addition of say 2mm
during the comparisons with theory would have disordered the frets on the diagram, and also
further sharpened fret 1 to QCM, or even TCM. In contrast, the procedure has proved to be
useful in making citterns. Explanations may be very low nuts, frets and actions, particularly on
the orpharion, and some old citterns may not have been fully adjusted. The nut shift might be
viewed as a wise initial adjustment in the correct direction, with a final effective amount less
than 2mm. A shift of 1mm would be similar to deviations seen on the diagram, and equivalent to
only about 3 cents. These variations are comparable with the predictions for light pressing on
the different courses, which would be accommodated by some average fret position. This matter
will be revisited in the analysis of Palmer’s frets, after the following sections on tuning theory.
The analysis of stretch-sharpening can be extended to all other wire-strung fretted instruments,
such as the orpharion, different sizes of cittern, etc. From the equations, the effects scale simply
with size so that, for example, the longer string lengths can use deeper scallops (Ref 17).
The Rose orpharion can be compared with the citterns. The relative rise of pitch was found to be
f /f = x/2x0, where xo = 4 f 2 L03/E. The value of L0 is increased from 430 to 600mm, and for
the same pitch g, midway between the 3rd and 4th courses, xo is increased by a factor of 2.7,
which reduces the pitch rise by the same factor. For the top course g, xo is increased further by a
factor of 4, and a total of 11. Conversely, the pitch rise for the 6th course G would be increased
by a factor of about 1.5, but then reduced by a factor of perhaps 2 from the increased elasticity
of the twisted wires needed on the 5th and 6th courses. These sensitivities would be reduced even
further by lower fret heights.
A modern guitar with metal strings is often said to be immune from gross stretch-sharpening on
account of the high string tensions. Here the equivalent factor is L0, which is probably about
700mm. This gives sharpening 0.23 times that for 430mm, so the effect is less than a quarter.
A final extreme application could be the various instruments of Indian classical music. Here the
fret heights are large and whole melodic phrases of many notes over a range of about an octave
can be produced after a single pluck. In this effect, known as meend, the string is also pulled
sideways, which requires great physical hardiness. It imitates vocal music and is particularly
expressive on larger instruments, such as veenas and the surbahar. Several instruments that now
have metal strings started many centuries ago with the familiar range of biological materials. It
is not easy to find details or dates of the changes within an unbroken, largely oral tradition.

MORE on TUNING and TEMPERAMENTS

It was seen above that cittern frets on real instruments predated Gerle, and perhaps all other
known written instructions. Finding some origin of cittern frets is therefore an intriguing
problem. In Ref 1, it appeared that Gerle’s fretting scheme for particular and also general
tempering evolved as a contrast out of Pythagorean tunings. After studying both the schemes
and real frets, another influence can be suggested, and this also has some relation to the
Pythagorean systems.
In Gerle’s scheme there are several Pythagorean or perfect intervals such as the fourth, fifth, and
major second. Others, including Dowland, Ganassi and Bermudo, used a perfect major third,
minor third, major and minor sixths and sevenths. However, no consistent scheme for a lute, or
even a keyboard, could use more than a few perfect intervals. In Ref 2 the starting point was a
simple demonstration that for a scale of twelve ‘perfect’ notes several intervals within the scale
show serious inconsistency and poor tuning. Attempts to set up so-called natural tuning or ‘just
intonation’ on a keyboard require several keys for the same nominal note (eg Refs 7, 9). In
addition, a lute would require different fret positions on adjacent strings, or staggered frets. This
is clearly impossible with simple gut frets, and scarcely practical with many small added
‘tastino’ frets. It might be more feasible with fixed frets on a cittern, and modern makers fit
some enharmonic alternatives on bandoras. Later in the cittern’s life van Nierop suggested a just
intonation scheme, which is examined below.
The main purpose here is to present some new general relations I have found between perfect
intervals and meantone tuning, which as shown in Refs 1 and 2 is the only consistent type of
tuning using simple frets. This has intrinsic interest and could have helped the cittern makers,
then Gerle and others. Table 1 in Ref 2 gave a list of ‘perfect’ intervals, and included
enharmonic alternatives. Initially it will be clearer and sufficient to list only the flats that are
appropriate for a basic scale on lutes and citterns. The frequency ratios are:
C Db D Eb E F Gb G Ab A Bb B C
1 16/15 9/8 6/5 5/4 4/3 64/45 3/2 8/5 5/3 16/9 15/8 2
It is useful to inquire about the origin of this ‘perfection’ that has so strongly influenced and
determined tuning systems. In renaissance polyphony only the octave and fifth were regarded as
perfect consonances, and this has a good basis in acoustics. The octave C has a difference of 1
or simply the initial C. The fifth G sounding with the initial C has a difference tone of ½, which
is C an octave lower. If slightly different frequencies are used for the fifth, octave or another
initial C then the difference tones are persistent beats sounding out of tune. This simple theory is
familiar, but for the other notes the arithmetic and physical reality become more tenuous.
Thirds and sixths were considered to be imperfect consonances. Later it was seen that a major
third E produces a difference of ¼, which sounds two octaves lower, and so came to be
recognized as perfect. The difference resulting from the minor third Eb is 1/5. This is technically
a seventeenth, or Ab two octaves and a major third, below the initial C, since 1 = 4 x 5/4 x 1/5.
This is a less obvious relation, both in sensation and numerically. A major triad contains a minor
third between E and G, in addition to the fifth and major third; and a minor triad contains the
major third Eb to G.
Similar relations are found for the sixths. For example, the major sixth A gives a difference of
2/3, which is F a fifth below the initial C, since 1 = 3/2 x 2/3. The minor sixth Ab has a
difference 3/5 which is a major sixth below C, since 1 = 5/3 x 3/5. Both these relations have
interesting symmetry. A special instance of the latter relation arose in Dowland’s fret
instructions (Ref 1).
Seconds, sevenths, chromatic intervals, and even fourths in two parts, were regarded as
dissonances. However, the difference between D and C is 1/8, which is three octaves lower.
The difference for Db is 1/15, three octaves and a seventh lower since 1 = 8 x 15/8 x 1/15. The
fourth F gives a difference of 1/3, an octave and a fifth lower since 1 = 2 x 3/2 x1/3. The
remaining notes Gb, Bb, B do not give differences with a perfect relation to C. For B the
difference is 7/8, and the prime 7 prevents a perfect relation to C, as seen later. The same occurs
for Bb with the difference 7/9, and for Gb with 19/45.
As seen below, an alternative perfect D is the minor tone 10/9 and this has a difference 1/9,
which is three octaves and a major tone lower, since 1 = 8 x 9/8 x 1/9. The other common
alternative is 9/5 for Bb with a difference 4/5, and a major third lower since 1 = 5/4 x 4/5,
showing the simple symmetry of the A and Ab above. (The symmetry 4/3 x 3/4 would derive
from 7/4 which is far from perfect, and then 5/6 from 11/6 etc). The successive differences can
be summarized as:
Db - B - N; Da - D - C; Eb - E - C; F - G - C; Gb - N; Ab - A - G - C; Bb - N; Bab - E - C.
Here N denotes a non-perfect result, and the subscript ‘a’ the alternative form. Similar relations
could be found for the sharps.
This lengthy treatment demonstrates that some perfect intervals are ‘more perfect than others’.
About half have immediate sensational and musical significance. The rest display a large desire
to continue the numerical simplicity by using only small prime numbers such as 2, 3, 5. This
gives the Pythagorean 9/8 and 16/9, in addition to the basic 3/2 and 4/3. Multiples of any one or
two primes cannot equal the third type, or even approximate it for only a few successive
intervals. The intervals based on 5 cannot be Pythagorean and so create several
incompatibilities. On the other hand, the presence of 5 in both numerator and denominator
produces a perfect fifth from major and minor thirds, 3/2 = 5/4 x 6/5.
The descending sequence of 3/2, 4/3, 5/4, 6/5 cannot continue with the prime 7. The ratio 7/6
would be a very small minor third and 8/7 a large tone. These were used in early Greek theory
but not in later times. Related notes appear in the harmonic series, such that 7/4 is a very flat Bb
and 11/8 is a very sharp F, both way off the diagram.
Another feature of the list of perfect intervals is a desire for symmetry, but this is not complete.
There are several perfect fifths and perfect thirds, and the ancient vain hope was that some
usable system could be found. Several of the simplest ratios cannot give rise to perfect intervals.
The deviations are large, generally by a syntonic comma, 81/80. For example, a rarely needed
Db to Ab is perfect, but a commonplace D to A is flat by a whole comma. The problems obtrude
all over the system, whereas in a consistent meantone the problems are pushed into unused
regions. The first known work on just intonation was by Ramis in 1482, followed by many
famous figures such as Agricola, Kepler, and much later by Mersenne in 1636 (eg Ref 7). No
proof was given that such tunings are impossible, probably because each problem could be
answered by some alternative form, yet never lead to a complete solution.
Despite these problems I have found some general relations between the natural intervals and
meantone, which are normally considered as completely different systems. These allow one to
classify perfect intervals and they probably paid some part in establishing general meantone
tunings, simply because there was nothing else apart from the pure Pythagorean system (a form
of reasoning known as abduction). These relations arose out of the work in Ref 1, where suitable
ratios were sought for constructing fret positions. One way of obtaining ratios took two exact
extremes and then found an intermediate case.
The clearest example of extremes was the major third E in QCM with the perfect 5/4, whereas a
major third in PYT is 81/64, from four perfect fifths (3/2)4. The interval in PYT is wider than
perfect by the 81/80 comma, so that 5/4 = (3/2)4 (80/81).
Also, in TCM the minor third Eb is the perfect 6/5, but in PYT it is the narrower 32/27, so that
6/5 = (4/3)3 (81/80) /2.
These two cases are familiar, but the method can be reversed to find and check all the perfect
intervals, and to assess any other candidates. It will be shown that every perfect interval can
define a particular temperament. Some traces of this will be found in Gerle’s scheme, and other
ancients may have used the approach.
For the seventh B the interval in PYT is 243/128, from five fifths (3/2)5. Reducing this by a
factor 80/81, as for E, gives 15/8. This derivation is a confirmation of the value taken as a
perfect interval. In addition it can be seen that the ratio 15/8 belongs to the system of 1/5th
comma meantone, since 15/8 = (3/2)5 (80/81) /4.
These three cases can also be inverted. This gives Ab with a perfect value of 8/5 in QCM, since
8/5 = (4/3)3 (81/80) /2.
Then A is a perfect 5/3 in TCM, since 5/3 = (3/2)3 (80/81) /2.
The most interesting case is that Db is a perfect 16/15 in 1/5th comma, since
16/15 = (4/3)5 (81/80) /4. This has very clear significance for fret 1, in the frets seen on citterns
and orpharions, and Gerle used the similar slightly smaller 33/31.
Three temperaments have now been defined by six perfect intervals, which could be used to
position six frets. The method can be extended to the remaining cases.
A perfect F# in SCM is 45/32, since 45/32 = (3/2)6 (80/81) /8.
The perfect Gb in SCM is 64/45, since 64/45 = (4/3)6 (81/80) /4.
Next we find a perfect C# of 135/128, and a perfect Cb of 256/135, both in 1/7th comma, then a
perfect G# of 405/256 and a perfect Fb of 512/405 both in 1/8th comma or ECM.
Further temperaments are 1/9th comma defined by a perfect D# and Bbb, 1/10th comma by a
perfect A# and Ebb, etc. It might be of interest that 1/11th comma, which is close to ET, involves
a perfect E# and Abb.
The remaining notes are D, Bb, G and F, which ancient initial schemes usually gave as
Pythagorean. However, real fixed frets have been seen above with tempered positions. The
present analysis can be extended to supply alternative ‘perfect’ notes, as mentioned above.
The major second D is the perfect Pythagorean 9/8 and the value in a tempered system would be
(3/2)2 (80/81) /2, which is 10/9. This is generally considered as the minor tone in a natural scale
so that a major third, such as C to E, contains the tones 9/8 and 10/9. Here it is defined formally
as another perfect interval characterizing ½ comma meantone. A complementary Bb is given by
(4/3)2 (81/80) = 9/5. In ancient attempts to construct just scales the initial D and Bb, based only
on the primes 2 and 3, result in deviations of a whole comma. Using one of the alternatives can
shift the problem around, but not remove it. This analysis reveals the correct status of these
intervals.
Extending the process to the perfect fifth G, the value in a tempered system would be
(3/2) (80/81) /2 = 40/27. This is known as a ‘wolf tone’, but it could be labelled as 1 comma
meantone. The complementary fourth F is given by (4/3) (81/80) = 27/20. These poor
alternatives, and even worse forms, actually appear in some theoretical just scales.
The perfect ratios for the flat enharmonics used on a lute are shown by a letter P on the diagrams
for Rose’s orpharion and the English cittern. Nearly all values lie below SCM, but the four
alternatives would be way off the diagram, so a list is also useful:
Note C Db D Eb E F Gb G Ab A Bb B C
Ratio 1 16/15 9/8 6/5 5/4 4/3 64/45 3/2 8/5 5/3 16/9 15/8 2
10/9 (27/20) (40/27) 9/5
Comma 1/5 0 1/3 1/4 0 1/6 0 1/4 1/3 0 1/5
1/2 (1) (1) 1/2
This shows how perfect intervals are related to Pythagorean intervals by tempering. The
relations for Db, B, Bb, Gb may be entirely new. In Ref 1 and above, the ratio 16/15 was seen as
a useful intermediate value for fret 1, but its exact definition of 1/5th comma had not been
recognized. The relations also give a new way of viewing D, Bb, F, G. An unexpected bonus is
that the relations in the list actually amount to a rigorous proof that natural or just intervals
cannot form a consistent tuning on any simply fretted instrument. A further task may be to prove
that all possible ratios based on the primes 2, 3, 5 cannot produce a consistent system on any
fixed note instrument. This seems a little ambitious since Mersenne, as a famous number
theorist, only presented possible reasons. (He is best known for the Mersenne primes Mp= 2p - 1,
where p is also prime; and to physicists for treating a vibrating string as a mass on a weightless
string before Newton had presented his mechanical theories.)
It is interesting to list perfect ratios and temperaments for some sharp enharmonics:
Note C C# D D# E E# F# F## G# A A# B C
Comma 1/7 0 1/9 1/4 1/11 1/6 1/13 1/4 1/3 1/10 1/5
Such a scale would not be an obvious choice for a lute, and there is no symmetry with the flats.
Some further relations can be deduced for the perfect intervals. Products of pairs such as Db and
C# are equal to D, and products of pairs such as Db and B are equal to 2 times D. Also the ratio
of all pairs such as Db to C#, ie diatonic to greater chromatic semitones, can be expressed as
2048/2025 = 211/3452 = 1.01136. This has some similarity to the two types of comma, and turns
out to be the rarely used ‘diaschisma’ (split apart). It was normally defined as the excess of three
octaves over four perfect fifths and two major thirds, or the excess of two diatonic semitones
over a major tone, (16/15)2/(9/8). There is also a ‘schisma’ which is the excess of a syntonic
comma over a diaschisma.
In Ref 2 I found that the ratio of flat to sharp enharmonics in ‘x comma meantone’ was given by
((4/3)(81/80)x )12/32. For the special case of 1/6th comma this is exactly equal to the above value
of the diaschisma for perfect intervals. Stronger tempering gives a larger ratio, and weaker
tempering a smaller value, which becomes 1 for ET. This abstruse theoretical measure of the
closeness of perfect intervals to SCM is shown below to have practical significance and use.
While the new relations between natural, Pythagorean and meantone tuning are the principal
firm results, they also suggest how the ancients could have used the natural basic intervals for
initial constructions of tempered meantone systems. For example, the ratio 16/15 was seen as
useful for a fret 1 in Ref 1, although that earlier work was more ambitious in seeking a further
construction with the same tempering, as Gerle had done. Next a fret 2 could use a reasonable
average of the two extreme ratios, then a fret 3 in TCM and a fret 4 in QCM would be good
starting points. Frets 5 and 7 could initially be left perfect, and fret 6 could have an exact SCM.
The higher symmetric frets could follow similarly. All of the constructions would be simple
with no difficult fractions. The initial positions could be adjusted by ear to give the basic cittern
fretting with its quite consistent tempering of the more difficult frets. Gerle’s scheme could have
been a slightly later refinement, particularly in the relation between frets 1 and 3 to produce a
strongly defined meantone. The average of all the different initial temperings would be about
1/7th comma, using simple untempered frets 2, 5, 7, 10. An average tempering of ¼ comma on
frets 2 and 10 would give an overall average near SCM. An extreme tempering of ½ comma on
frets 5 and 7 would put the average near QCM. In a similar way, the above scale containing
sharps might be a reasonable start for ET, since most notes are less tempered than SCM, and the
average is about 1/10th comma. These tuning schemes might be grouped as Coakley4, following
the other examples constructed in an ancient style.
It is important to recall that the earliest writing on meantone by Aron in 1523 was only a decade
before Gerle. This was not a mathematical division of the comma, but an instruction to narrow
four successive fifths to give a pure third, which is still used for tuning keyboards and harps to
QCM. This type of method can also be used today to tune open strings on fretted instruments
and then position frets using various checks. Initial perfect thirds and fifths could be changed to
any tempered system between PYT and QCM. As noted above, on the diagram a simple average
would be ECM, which is also midway between ET and SCM. The broad range can be given
greater definition by the evidence of real frets on citterns and later written fret constructions
from the lute theorists. The new suggestion of starting from a set of many perfect intervals
defines more precisely the SCM found on real citterns. This approach to tuning is a plausible
ancient theoretical basis before Gerle’s writing, and it is even possible his scheme was a
variation on this idea. Such an initial set-up requires far less adjustment than many other
schemes, including Gerle’s. It was found in Ref 1 that early Pythagorean schemes involved
laborious constructions that were prone to error. This might lead to some accidental tempering,
but generally they would need much adjusting so that other methods were necessary.
In relation to a later section on baroque tunings, I found that Mace mentioned positioning frets
on citterns and lutes (Ref 19, p70). He said there was a ‘mathematical rule for bandores,
auferions and citterns’ with wire strings, but that the rule does not hold exactly or always for gut
strings on account of their frequent falseness. He advised tuning by ear to perfect intervals of a
major third, fifth and octave, firstly for the open strings, and then to place the frets. As seen
above, this could lead to any sort of meantone. A tantalizing fear that we may have lost a special
ancient rule for citterns etc may be quelled if he was simply referring to the standard
constructions for his three perfect intervals, which then needed adjusting before applying to the
cittern. This is likely because there is no other explicit maths in his book. The unusual
‘auferions’ has little relation to the Greek roots and may indicate a loss of familiarity with an
extinct instrument. A later ‘orpheoreons’ from Curt Sachs is half correct. When young, Newton
had calculated logarithms of perfect intervals - two centuries before cents (see Ref 9). A
logarithm converts physical ratios into our linear perception of intervals. As a singer at Trinity
College, Mace might have missed an opportunity to rekindle Newton’s interest in tunings. The
latter’s work on optics in the 1670s produced the famous analogy between spectral colours and
the notes of a scale.
Some further musical relations between intervals can be mentioned. In Ref 2 it was seen that the
sharpening of the fourth and major third by x and y proportions of a comma can be expressed as:
(4/3)4 (cx)4 (5/4) cy = 4 so that y + 4x = 1, or y = 1 - 4x
The simple case of QCM has perfect thirds with y = 0 and x = ¼. For TCM with x = 1/3 the
minor third is perfect, since a major sixth is a fourth plus a major third with x + y = 0. The major
third is flattened by y = - 1/3, which would not be helpful.
The case of x = 1/5 is usually characterized by equal tempering for the fourth and major third
since y = 1/5. It was noted in Ref 2 that finding a neat musical relation can be less simple than
using the full theory. The present finding of the perfect diatonic semitone, C to Db, can be
related to the sequence C-Db---Ab--C, since the fifth is narrow, and the third wide, each by 1/5th
comma.
It was found above that in SCM the interval C to Gb was perfect. After some thought, this can be
related to the sequence C-Gb--Bb---F---C, which covers two octaves. Here y = 1/3, so the third is
wide by 1/3rd comma, and the two fifths are each narrow by 1/6th comma. It can also be noted
that in ET the value of y is about 0.64, so thirds are almost twice as sharp as in SCM, which is
about midway between QCM and ET, as seen on the diagram. The milder ECM has x = 1/8 and
y = 1/2, which is the midpoint of QCM and PYT, and almost midway between SCM and ET.
Other temperings might be represented as similar musical sequences of intervals.
Van Nierop’s just intonation
The difficulties of just tuning on keyboards, and its practical impossibilities on lutes, have been
outlined above. In 1659, after much of the music that interests cittern players, van Nierop
described a limited system of frets. This produces some perfectly tuned chords beneath each
note on the top string of a cittern in French tuning. I have not seen the original, nor Barbour’s
paper on it, but the chords up to fret 6 are listed in Ref 9. The reproduction of an original
diagram of the frets was illegible, so I needed to calculate the positions and pitches. As can be
expected from the last section, this scheme demonstrates that the frets on old citterns cannot be a
form of just intonation. It is also an introduction to the more difficult case of a lute. On a
positive and surprisingly practical note, I also found that a ‘best attempt’ to set up just intonation
can actually produce almost exact SCM frets.
The open notes, e, d, g, a, were listed a fifth lower as A, G, C, D. This will be retained, to help
any future reference, and it also matches the C scale listed above. The intervals between open
strings were C-D 9/8; C-G 3/2; and hence D-G 4/3; then G-A 10/9, and hence C-A 5/3. These
perfect ratios can be seen in the just scales above, and notably the alternative 10/9. However, the
interval D-A is 40/27, a whole comma flatter than a perfect fifth, so that a simple chord with
open 2nd and 4th courses is not possible. This type of problem occurs in all attempts at just
tuning, even on a standard keyboard. The various chords used by van Nierop can be listed by
fret letters (a - open, b - fret 1, etc), and course numbers (1 for the top course A, and 4 for D).
For an open 1st course his only chord was (a1, c2, a3, d4) or (A, A, C, F).
For fret 1 the only chord was (b1, a2, c3, a4) or (Bb, G, D, D), where the open D is compatible.
For fret 2 the only chord was (c1, a2, c3, a4) or (B, G, D, D).
For fret 3 two chords (d1, a2, a3, c4) and (d1, c2, a3, d4), or (C, G, C, E) and (C, A, C, F).
Fret 4 was missed, and the use of e for fret 5 and f for fret 6 was initially puzzling.
For fret 5 two chords (f1, a2, c3, a4) and (f1, d2, f3, a4), or (D, G, D, D) and (D, Bb, F, D).
For fret 6 two chords (g1, f2, a3, f4) and (f4, a2, d3, f4), or (Eb, C, C, G) and (Eb, G, Eb, G).
From these chords and open tunings, the required fret positions can be found. Different sets of
positions are needed on each course, so that the frets are staggered. These can be listed in terms
of frequency ratios on each course:
Fret 1 2 3 4 5 6 7
Course 1 27/25 9/8 6/5 miss 27/20 36/25 3/2
2 10/9 6/5 4/3 3/2
3 9/8 6/5 27/20 3/2
4 10/9 32/27 4/3 40/27
The ratios have been offset slightly to indicate the relative positions. The difference of position
between adjacent courses is a whole comma, 81/80, which is about 7mm near the nut or 4mm
near an octave fret. There is some degree of pattern, and this corresponds with the original
diagram. Higher frets were also surmised but are not given here, except for the clearer seventh
frets, since the original required chords were not known. The pitches at each fret are found to be:
Fret 0 1 2 3 4 5 6 7
Course 1 5/3 9/5 15/8 1 9/8 6/5 5/4
2 3/2 5/3 9/5 1 9/8
3 1 9/8 6/5 27/20 3/2
4 9/8 5/4 4/3 3/2 5/3
The resulting notes are C 1; D 9/8; Eb 6/5; E 5/4; F 4/3 and 27/20; G 3/2; A 5/3; Bb 9/5; B 15/8.
Obtaining perfect tuning of only a few notes for a few basic chords has required a large
departure from simple frets. Most notes can be seen in the scale listed in the last section, but in
addition, the note F requires two alternative pitches 4/3 and 27/20. This is purely a result of the
internal inconsistency of the just scale, and not the fretting of different courses. The chord F A C
requires the pitches 4/3, 5/3, 2 for the perfect ratios 1, 5/4, 3/2. The chord Bb D F requires the
pitches 9/5, 9/4, 27/20 for the same perfect ratios. This illustrates the central problem of just
tuning. While there are many perfect fifths and thirds in the scale, there are also several large
discrepancies for common notes. In contrast, the problems in a meantone can be confined to an
unused region.
The problem in the just scale first arises with the notes D and Bb, which were shown above to
have simple alternatives. Early theorists often gave their name to a single changed note, but the
basic intervals had all been known to the Greeks. The scale above was chosen for its symmetry,
but I cannot trace a previous instance. With 10/9 for D it is called Mersenne’s spinet tuning no1,
a 9/5 for Bb gives his lute tuning no2, and both alternatives give his lute tuning no1. Other scales
had deviations of two commas, and many included sharps, as in Mersenne’s spinet tuning no2.
Whichever scale one chooses the central problem pops up obtrusively, like a mole in a perfect
lawn, whereas meantone keeps him in the surrounding beds. For any failed system another
alternative note can be suggested, so the main problem for proofs about just intonation would be
listing all possibilities. Some of these might have practical uses in special situations.
I have tried setting up just frets for a lute, but as might be expected for the span of an extra fifth
between the open tunings, there are more differences and relatively fewer chords than for the
French cittern. Transforming a just scale by a fifth between open strings generates alternative
forms of D and Bb, and further fifths produce more changes. In addition each fifth alters an
enharmonic, as in meantone. From a ‘best attempt’ at a complete tuning the staggered frets can
be averaged to simple frets. Very interestingly, I have found this gives an almost exact
description of SCM, which would apply equally to citterns and lutes. This takes further account
of the preferred enharmonics on different courses, and extends the idea on early possible
constructions for meantone. The theoretical basis is the relation involving the diaschisma, which
I found above between SCM and perfect intervals. All this shows that SCM is actually a
significant system, rather than merely a popular useful choice where the major thirds are
tempered twice as strongly as the fifths. (It would take several pages and drawings to present
this, but the procedure may be apprehended from the above discussion.)
Another possibility would be to use staggered frets at the nut, rather than simple equal lengths
for open strings. This could also be applied to van Nierop’s scheme. On a practical note, simple
gut frets would need many tastino frets, or a change to fixed frets. This brings to attention a
serious objection to staggered frets on a cittern, and some tastino lute frets. In addition to
difficulties with single fingers, the common technique of barred fingering across courses would
be greatly hampered, or impossible. Nevertheless, if not taken too seriously, thinking about just
intonation on a lute can be more interesting and useful than many other games and puzzles.
The next section on baroque lute tunings could be missed by readers only interested in wire-
strung instruments. Those more interested in lutes will need the background from earlier
sections. The FINAL REMARKS section refers to both types of instrument.

BAROQUE LUTE TUNING and TEMPERAMENTS

The criteria found above involving the tuning of open strings of any fretted instrument were
stimulated by moving on to citterns after the renaissance lute. In turn this has prompted some
new thought on the development of the baroque lute. It will be useful to show the interrelations
between all these fretted instruments, rather than produce several accounts requiring much
repetition. There is also ready reference to the new criteria derived above, mainly at the
beginning of CITTERNS. These considerations outweigh a simple preference for many separate
papers on different instruments.
D minor tuning and Vieil ton
The best approach is from the final tuning of the baroque lute, for technical rather than historical
reasons. Then one can understand the earlier intermediate or transitional tunings that were used
during the evolution from the renaissance lute. The final tuning of the six main fretted strings is
A d f a d f, known as D minor tuning. This is spanned by 4 successive fifths F, C, G, D, A,
which are exactly the same as in the renaissance tuning with G c f a d g, known as ‘vieil
ton’. The D minor tuning would therefore entail the same constraints for a meantone type of
tuning that were deduced above for the renaissance lute and cittern. However, it is almost certain
that the standard baroque lute used an ET tuning. Modern players would naturally assume this,
and be dismayed by mention of ‘meantone’. One can understand that ET would be necessary for
the wide range of keys that were in regular baroque use. Pieces were written in keys from Ab
major and F minor, through to A major and F# minor, and each composition could also require
slightly more remote chords. This would not have been possible with a meantone such as SCM,
or a milder ECM.
In contrast, pieces for renaissance lute had used a narrower range of keys from Eb to perhaps D
major, and C to perhaps A minor. Keys such as Bb minor and A major are rare and later
compositions would be expected to need a tempering close to ET. For example, Db major and Bb
minor require a Db with fret 4 on the A strings, whereas a meantone has a C# that cannot be
altered without harming all the other courses. Conversely, A major and F# minor require G# with
fret 3 on the F strings, in addition to the F# and C# on fret 1. The use of individual tastino frets
for both alternative notes in the centre of the fingerboard would be difficult and also hamper
barred fingering. Searching for some examples, Dowland’s rare use of F minor in ‘Melancholy
Galliard’ would be possible with meantone and only one bar needs the above Db. Laurencini’s
many pieces in Bb minor look in great need of ET, but the thick chords may sound acceptably
dark in SCM, with flat minor thirds. (This may have some relation to Rose’s flat fret 3.) In
contrast, his pieces veering to the sharper keys may sound generally poor, since the ‘flat’ type
enharmonics are sharp, although one might develop a taste for some contexts. This initial
analysis produces a clear distinction between standard D minor baroque usage and most
renaissance music. A key area of interest is the original French exploration of new tunings, or
‘accords nouveaux’, which began around 1600.
Transitional tunings
A totally new musical style apparently arose from the brain of Mesangeau, followed by several
others such as Chancy and Bouvier. Virtually all their known pieces are contained in the CNRS
editions (Ref 20), and a striking fact is the very narrow range of keys and modes. One reason for
this may be the constraints of fingering patterns with the new open strings. Another limiting
factor could be the use of general meantone systems other than ET. A use of ET would only be
affected by the fingerings, which could allow a wider range of keys.
The main tunings used over the brief transitional period can be listed as:
Vieil ton G c f a d g 4 fifths
French Flat G c f ab c eb 5 fifths
Mesangeau G c f a c eb 6 fifths
Gautier G c f a c e 5 fifths
Sharp tuning G c f a c f 4 fifths
The 7 and 8 courses, tuned to F, E or Eb, were sometimes used on low frets and can match the
th th

higher courses. The notion of ‘flatness’ refers to the 1st and 3rd courses, and other labels for the
tunings are also in use. The CNRS notation uses a 6th course G rather than A, and this is helpful
to show the development from vieil ton. For comparison, the D minor tuning could be written
as:
D minor G c eb g c eb 4 fifths
In the new tunings the top two courses are relatively lower by about two whole tones. The old
two-octave range was redistributed over eight courses, as explained in Ref 16. In contrast with
vieil ton, the upper strings are separated by thirds, the centre by fourths, then three close basses.
D minor tuning had greater changes in courses 3 and 4, with an intermediate pattern of intervals.
These tunings are more radical than scordatura used later on violins. This was for occasional
chords with alternative colour in a few pieces, and far from a permanent change in the largely
monophonic violin. For the lutenists each new tuning was effectively a different instrument,
which they explored remarkably rapidly.
The accepted reason for these many changes of the open strings is a search for sonorities
pleasing for the new musical style, in conjunction with new designs of the lute itself. When one
changes between the above tunings, perhaps by only a semitone on one string, there can be a
different feel. Similar effects can be found by slightly altering the overall pitch of a given lute.
While the general idea may be seen, an unimaginably fine degree of understanding the lute’s
acoustics would be needed for a full explanation (Ref 21). The new idea here involves the
relative tuning of the open strings and it is possible to be more certain.
Allowed keys
The above open tunings also list the number of successive fifths needed to generate the open
strings. For the relatively popular French Flat tuning the G string might have the same scale of
notes and frets as vieil ton:
G Ab A Bb B C Db D Eb E F F# G
and 5 successive fourths transform the notes to an Ab string with
Ab Bbb Bb Cb C Db Ebb Eb Fb F Gb G Ab
The notes in common are G, Ab, Bb, C, Db, Eb, F, which is only seven out of twelve. Each fourth
changes the enharmonic of the leading note or fret 11, which then appears on fret 4, eg F# to Gb.
Matters are not improved if the Ab string started with fewer double flats. For the following
analysis, readers may want to write out the transformations on all the different strings.
This situation for French Flat tuning can be compared with the span of only 4 fourths from A to
F on the renaissance lute. From the above G string the F string has
F Gb G Ab A Bb Cb C Db D Eb E F
and the A string has
A Bb B C C# D Eb E F F# G G# A
There are now eight common notes- F, G, A, Bb, C, D, Eb, E, and importantly the differences are
subtly shifted and turned to advantage. As is well known and explained above, the Gb is a
problem but Cb less trouble, while C# on the A string and F# on the D string become very useful.
This comparison shows that a general meantone would be able to handle a more restricted range
of keys for a French Flat tuning than for vieil ton. There may also be restrictions due to different
fingering patterns. This would entail a long exercise, even for the familiar vieil ton, which has
more uniform intervals between open strings. There is no simple rule other than a lengthy
enumeration of all possible fingerings, but see later.
Mesangeau’s tuning has one further seemingly small change, of a third course a instead of ab.
The open strings now have a span of 6 fifths, from Eb to A. There are only six common notes,
and the problem with enharmonics has been increased, with sharps on the ‘a’ string and double
flats on the eb string. There would also be changes in fingering patterns.
Gautier’s tuning has a further sharpening of the 1st course from eb to e. This reverts to a span of
5 fifths, but now between F and E. This is like vieil ton with another fifth above the A.
The further example of a ‘sharp’ tuning has exactly the same span as vieil ton, but there may be
few surviving French pieces. Before the transitional tunings the 6th course G had been lowered
to an F. Similar ‘avalée’ tunings centred on a single common chord left very little music.
Musical pieces
It is worth giving full lists of pieces by the early baroque composers according to key and
tuning, which I have painstakingly assembled from the CNRS publications.
Mesangeau: Vieil ton 1 in F, 2 in F minor, 3 in G minor
French Flat 5 in Ab, 15 in F minor
Mesangeau 9 in Bb, 1 in C minor
Gautier 6 in F, 3 in C, , 1 in A minor
D minor 1 in D minor (a sarabande finely recorded by Rolf Lislevand)
Chancy: French Flat 6 in Ab
Gautier 6 in C
Bouvier: French Flat 5 + 6 in Ab, 6 + 3 in F minor ( + indicates fragments)
Mesangeau 7 in Bb
D minor 5 + 2 in D minor
Belleville: Vieil ton 2 in C, 3 in C minor, 1 + 2 in G minor
French Flat 3 in Ab, 2 + 1 in F minor
Dubuisson: Gautier 2 in C, 1 in A minor
b
Chevalier: French Flat 4 in A
Gautier 5 in C, 4 in D minor, 1 in G minor
It is interesting that all surviving pieces by the famous Ennemond Gautier, who lived from 1575
to 1651, appear to use the D minor tuning but none of the older tunings. (Gautier tuning is
named after an unrelated composer.) However, Dufaut with dates between about 1600 and 1670
is represented by about 30 pieces in transitional tunings, which are rarely heard, in addition to
over 100 pieces for D minor tuning. His keys for French Flat and Gautier tuning are the same as
above, but no pieces use Mesangeau’s tuning. For D minor tuning his keys range more widely
from Bb to A major, and from C to E minor. One or two of his pieces used the minor keys of B
and F#. The latter key became a favourite of later composers, especially Gallot.
Summarizing the detailed lists, the only keys for the many pieces with French Flat tuning are in
Ab and F minor. For Mesangeau’s tuning all the pieces are by him or Bouvier, and are in Bb,
with a single exception in C minor. Also, these two composers are the only ones in the above list
using the new D minor tuning. For Gautier tuning, the fewer pieces were in C, F, and the minor
keys of A, D, G.
The open strings for these three tunings become sharper, and so correlate with the keys. French
Flat and Mesangeau each mainly use a single key, whereas the wider range for Gautier tuning
may reflect its closeness to vieil ton. All this would have resulted from a mixture of convention,
changing taste, available fingerings, and acceptable temperaments. For further insight it is useful
to examine actual pieces in tablature.
For French Flat tuning, fret 1 requires Ab, Db, Gb on the G, C, F courses, and this is consistent
with the scales listed above. The 1st course normally uses only frets 0, 2, 4, 5, 7, and some pieces
are ‘sans chanterelle’. An E is occasionally needed instead of Fb, and an A instead of Bbb. Fret 2
is fine for all courses, but fret 3 sometimes needs B on the 3rd course, instead of Cb. The higher
frets are satisfactory, apart from the dilemma over fret 6. Also fret 8 might require B and E
rather than flats, but this is not consistent with an Ab on the 2nd course. For any lute, the higher
frets are not needed on lower courses, and over a quarter of available notes are unused. Frets
could be moved to suit the higher courses, unless adjacent courses prefer the initial enharmonic.
The relatively sharp C course between two flat courses Eb and Ab might be a slight hindrance.
This analysis demonstrates similar but greater problems than with vieil ton. It might be
concluded that the difficulties would force the ancients towards an equal temperament. This
could also have simplified further exploration of other open tunings. However, it can also be
seen that a greater tempering, such as ECM or even SCM, would be suitable for their keys of Ab
and F minor, and a step to either side. Ancient searches for improved sonorities could also
benefit from a purer meantone that had been used for vieil ton.
Regarding convention and taste, two of Mesangeau’s pieces in F minor for the earlier vieil ton
have chords with a bass F, an Ab from fret 1 on the G course, then an open c and f. This sonority
was extended in the new French Flat tuning by open ab and c courses. To what extent the keys
were dictated by fingering patterns might need long practical trial and even composition.
It is also notable that as late as 1676 Mace produced several suites in different keys for French
Flat tuning (Ref 19). Allowing for his higher pitch standard with a 6th course B, the equivalent
keys were Eb, Ab, Db, and minor keys of G, C, F, Bb. These are a step either side of the original
French keys. However, the pieces are simpler and the overall range is only slightly wider than
the French pieces. This also indicates that the originals were not limited by fingerings. One
might expect that Mace was using ET at this late date, and the drawing of his peculiar double
lute does not contradict this possibility. However, his initial placing of frets to perfect intervals,
as described above, could lead to any consistent meantone. For 1676 much of his work was old-
fashioned, and he left only one suite using D minor tuning, and in the same key.
For Mesangeau’s tuning, virtually all the pieces are in Bb, and the above scale can be assessed
for use in a general meantone. Fret 1 produces the useful Bb on the 3rd course, but also the
unwelcome Fb and Gb on the 1st and 4th courses, where E and F# are sometimes needed. Fret 3
with a Gb on the 1st course is unused. Bouvier’s pieces use the Bb and Eb on the 2nd and 3rd
courses, but little above fret 4. The use of a meantone would be possible and an enhancement for
his relatively simple, and beautiful, use of naturals and two flats. However, the appearance of
three sharps on the A string surrounded by many extreme flats is not helpful for further keys,
which is a great difference from vieil ton. Mesangeau’s frequent use of higher frets on upper
courses would need to alter some of the flat enharmonics. He also made more use of lower
courses. His unusual and original allemande in C minor may indeed be ‘an exception that proves
a rule’, since it ingeniously uses almost all the required naturals and flats, while avoiding the
many sharps and double flats. In detail, the 1st course correctly uses all frets up to 7, but misses
1 and 3, while fret 6 could be moved to a true A. The 2nd course is fine, using frets 2 to 5, and
the 3rd course is fine with frets 1 to 3. The 4th course is fine with frets 2 to 5, but in a single
instance needs an F# on fret 1. The 5th course is fine, using frets 2 to 4. The 6th course most
remarkably uses all frets from 1 to 5 with no problem. This striking piece shows that a general
meantone on the most restrictive open tuning is indeed possible. Many easier keys and less
demanding uses of frets could be mishandled in a meantone. Additional examples of Mesangeau
tuning in the key of Bb, and consistent with a general meantone, are the three duets by William
Lawes, and several pieces by Pierre Gaultier. The latter’s works use ‘accords nouveaux’ but
have several stylistic differences from the other composers, and are not found in French sources.
Detailed examination may be of interest, and it would be significant if any were incompatible
with a meantone.
For the Gautier tuning five keys were used in the above pieces, which could involve much
checking. A shorter approach can note that the main difference from vieil ton is one extra fifth
E. This introduces a new D# at fret 11 on the 1st course, which is no problem. The C# on fret 9 is
unused, while the F# and G# have correct uses. Pieces commonly miss most of the sharps on the
1st course. Similar reasoning can be applied to the more remote eb 1st course in Mesangeau
tuning, but the French Flat tuning has two open tunings different from vieil ton.
Early D minor tuning
The above evidence points to considerable influence of temperaments and fingering patterns on
the keys available to composers exploring ‘accords nouveaux’. Those still using vieil ton had
access to a wider range of keys with a general meantone, but some would have moved nearer to
ET for one or two more adventurous keys, as discussed above. After only a few decades the
pioneers settled on the D minor tuning, and this proved surprisingly amenable to fingerings in a
wide range of keys. These final open strings have further differences from the previous
transitional tunings, but are spanned by exactly the same four fifths as in vieil ton, F to A, and
courses 2 to 4 are identical. On this basis, the earlier compositions could have used a meantone
tuning, but a move to ET would be required for the much wider range of keys coming into the
baroque repertoire. For example, Ennemond Gautier died in 1651, and nearly all his 80 pieces
are in the minor keys of D, G, C. Only about 15 pieces are in D and A major, and F# minor,
which would need better sharps. Although D minor tuning became standard there were rare
variants such as Mercure’s low Bb instead of A; ton de trompette with a further bb; and D major
tuning with two F sharps. These would have been intended for handling flat or sharp keys before
ET was chosen.
However, a closer examination of the open courses and examples of tablature suggests that ET
would have been necessary much sooner after adopting D minor tuning. The two extreme fifths,
F and A, each appear on two courses, with no intermediate C and G courses that were used for
the earlier tunings. Problems could arise from the F courses with flats placed next to two courses
with more sharps. For scales on the F and A courses as listed above, the D scale would be
D Eb E F F# G Ab A Bb B C C# D
The scales for all the open strings F, C, G, D, A have now been listed. This covers all the fretted
courses for D minor tuning and vieil ton.
It might be thought that a melodic top course in f  would want the common sharps and flats,
such as F#, Bb, C#, Eb, and either Ab or G#. However, the sharps would produce notes like D# and
A# on the D and A courses, inconsistent with the required flats. In tablatures the problem was
often avoided where possible, by using a 2nd course fret 4, which is a true F#. This allows fret 1
to give Eb and Bb. This may have been less possible for the 4th course, where fret 1 was often
written for a true F#.
A similar effect would occur for a G# on fret 3, which would produce E# and B# rather than the
generally required F and C. In tablatures this was sometimes solved by a 2nd course fret 6, but
there are also many cases needing a true G# for fret 3 on the 2nd and 5th courses. Parallel types of
behaviour are found on higher frets.
For vieil ton, this alternative use of notes on lower courses was generally caused by the normal
course being used on higher frets for other notes. This can lead to many barred fingerings for
closely spaced chords, often on fret 3, particularly for intabulations of all the notes in a vocal
original. (This was demonstrated personally while making, then thinning out and modifying the
original ‘De mon triste desplaisir’, before obtaining the version by da Milano.) This cause is also
found in the baroque pieces, but chords are usually sparser. Some uses of lower courses were
not forced in this way, which may indicate use of a meantone and an awareness of the tuning
problems. In other cases a lower course was used for phrasing; for convenient closeness to other
frets in use; and also to double an upper course, often involving an ornament.
This analysis indicates that D minor tuning with a general meantone would produce greater
problems than vieil ton with the same span of fifths. This would be purely a result of the greater
number of the extreme open tunings F and A, their position among the courses, and their
musical use. With a full use of the alternative fingerings it might be thought that the same range
of keys available to vieil ton would be approached. However, this ideal situation would be
hampered by a restricted availability of notes. With a less frequent use of alternative frets, as
seen in the tablatures, the range of keys with correct enharmonics would be further limited. No
use of alternatives might reduce the range of keys down to about C to Eb major, and D to C
minor. Interestingly, this contains the larger group of E Gautier’s early pieces noted above, but
excludes the small group of sharper keys. An exact analysis would be lengthy, but players may
appreciate the problem from this more general approach.
The written evidence of tablatures shows the same frets frequently serving for alternative
enharmonics over the full range of baroque keys. The above observations therefore show that a
single general meantone fretting was not possible. There might be two solutions: either ET was
adopted soon after D minor tuning, or different meantone frettings were used for different
ranges of key. Some possible different enharmonic fret positions can be seen from examining
tablatures. Pieces in the more remote sharp minor keys reveal this most clearly. The notes F#,
D#, A# are needed on fret 1, and G#, E#, B# on fret 3, with similar requirements on higher frets.
Inspection of the tablature suggests that many, and perhaps all, pieces in E, B, F# minor could
actually be played in a general meantone by putting frets such as 1 and 3 in the ‘sharp’ position.
The values can be found from the diagram and Ref 2. Conversely, pieces in the flatter keys can
still use the normal ‘flat’ fret positions.
This would not be a complete solution since some frets cannot have compatible notes on
different courses. An example is an A on fret 4 for central keys, together with Db and Gb on the
other courses for F minor and beyond. This was discussed above for vieil ton, but the problem is
doubled with two main courses in F and two in A. For ‘sharp’ alternatives an A# on fret 5
produces F## and C##, rather than G and D needed for central keys. However, higher frets tend to
have more sharps even with the normal ‘flat’ frets, and so need less change. For example, a B
rather than Cb on fret 6 produces G# and D#. Higher frets are also used less, particularly on lower
courses, and can be set to suit the upper courses. Changing a fret to a different enharmonic is far
more complicated than simply retuning basses to suit keys.
For all these reasons, together with the convenience of one set of frets rather than three, it is
highly likely that ET was an early widespread practical solution for pieces using D minor tuning
and adopted a while before 1650. Paintings examined below also indicate this use of ET. Further
analysis might need to include hundreds of compositions and deduce their dates. This would be
an enormous task, but some readers may like to use the method. For now, there seems to be a
good correlation between the dates of the main composers and a generally increasing use of
more remote keys. Some approximate final dates for their pieces are: Mesangeau 1640, many
‘transitional’ exponents and also E Gautier 1650, Denis Gautier and Dufaut 1670, Gallot 1690,
Mouton 1700. Life spans were about 70 years, but many dates are not known exactly, and others
have been rounded off.
Nevertheless, an adventurous player might like to try certain pieces for D minor tuning with an
appropriate meantone. It might be thought that pieces using vieil ton in remote keys could also
use alternative meantone frets, but most would still require the incompatible notes and hence
ET. The newly recognized differences between vieil ton, transitional tunings and D minor tuning
are striking.
A possible role for secondary tastino frets can be examined. Writing in 1640, the keyboard
maker Jean Denis complained that a particular harpsichord was being tuned like a lute with
more equal semitones (eg Ref 9). He suggested staggered ivory frets, but there is no evidence of
their adoption. This may be further support for ET frets on D minor lutes by this time. It might
also appear to be an addition to the sparse earlier evidence for tastino frets. However, his ivory
frets were intended to replace simple ET-like frets to produce a more perfect tuning. This is
closer to the current notions of just intonation, rather than any earlier alteration of a few notes in
a general meantone. From the above analyses of just tuning one can see that staggered frets
would actually be far from a useful solution. It is also important to note that this late evidence is
no indication that renaissance lutes many decades earlier had ET frets.
Sympathetic Resonances
The final sonority that the ancients found with their D minor tuning is understandable for music
in the same key, with several open strings and many sympathetic resonances. These can be
identified by strongly sounding any string, then damping it, and testing for the many pitches
induced on other strings. The series of harmonics for each string is: 1st or fundamental, 2nd or
octave, 3rd or twelfth, 4th or double octave, 5th or seventeenth etc. It is important to note that the
3rd is not the fifth, and the 5th, which may be practically inaudible, is not the third. For example,
the bass D sounds D, d, a, d, f# etc. These pitches can be induced in another string only if they
are also harmonics of the second string. Examples are d in the 5th course d, and d in the 7th
course G. The entire system is easy to set up but takes several pages. However, there are several
features worth mentioning in addition to a general picture.
For the D minor tuning the main induced pitches concern the notes D, F, A, which is not
surprising. The string resonances are concentrated in this region. There is also a C, c, g series
from the C bass and octave, the series F, f, c from the F bass course, and G, g, d, but these have
less influence than the D series. Also, fretted notes will have different series which can both
drive and be induced, but only while fretted. In contrast, while certain notes may be deliberately
damped during playing it would be unusual to damp unplayed open strings which can therefore
strengthen the resonance around D, F, A.
Ancient use of the D minor tuning for other keys depended on finding fingerings on the same
open strings, but also required open basses to be retuned to the new key. The fingerings would
make less use of the open strings, and different colourings would arise from the new basses. For
flatter keys, the open strings and fretted strings, on the basis of their harmonic series, can be
expected to produce similar but weaker resonances. Basses in E or Eb are not highly involved,
but a further low Bb would be effective with F and Eb strings.
In strong contrast, for sharper keys leading up to F# minor, the altered basses would be G#, F#,
C#, which completely removes three of the main harmonic series G, F, C. This leaves only the D
series, but virtually no resonance could be induced by open higher strings. It can be expected
that resonances would depend mainly on driving by fretted notes, including some barred
fingerings. The resonances for these sharper keys would change in closer sympathy with the
plucked notes, and produce a lighter clearer, less muddled sound. The main fretted notes, with
fewer open notes, would also be better defined but less bright.
These experiments, both real and in thought, may provide new explanations for practical
players. Interestingly, resonances can be expected to be almost as effective for ET as some purer
meantone tuning since the main pitches arise from common naturals related by octaves, and
fifths that do not change much with tempering. It would have been helpful in a search for
sonority to find that an early change to ET for D minor tuning would not have completely
destroyed the string resonances. The analysis can be extended to vieil ton, where there is more
variety in the upper courses, and more complexity when combined with four basses on a 10
course lute. The effects for vieil ton are not concentrated around only a few harmonic series. It is
likely the ancients first noted the possibilities with the 7 to 10 course lutes. Then the various
‘accords nouveaux’ would have different complexities, since the resonances are related to the
open strings and the span of fifths between them. Finally D minor tuning would have produced
simpler effects, at first stronger but muddled, then clearer in the sharper keys.
With ET as a standard for D minor tuning it may be wondered why more remote keys were not
used. Regarding the string resonances, F# minor already achieved the clarity arising from fretted
notes rather than open strings. From the last section on temperaments, the keys eventually used
for ET appeared to terminate at around F minor, and F# minor possible with a ‘sharp’ meantone.
This would have been caused by different factors such as fingering patterns, avoidance of
double sharps and flats, and some correspondence with the restrictions on other instruments. The
complications of fingering patterns can be approached through the frequently required tonic
triads. The lowest frets needed on the main courses can be summarized as:
Ab Eb Bb F C G D A E B F# C# G#
F 3 2(1) 0(5) 0 4(3) 2 2 1(0) 4(3) 8(7) 3(2) 1 5(4) 3 3
D 1 1 0(5) x 3 2(1) 0 0 2 2 2 1(0) 4 3(2) 1
A 3(2) 1 1(4) 0 3 3 2(1) 0 4 0 2 2 4(0) 4 3(2)
This pattern repeats for three more courses, and the brackets contain the minor third. There is a
clear picture for the remote keys. The minor keys of E and F# often use barred chords, B is also
reasonable, but C# and G#, or Db and Ab, are difficult. These keys also have progressively fewer
naturals that can use open strings. For central keys, possible problems seen in the list, as for A
and even C, are offset by a greater use of open strings. For flat minor keys F has a barred chord,
but Bb is difficult. While Eb major was used, the minor key with a bar on fret 1 was overlooked,
probably because pieces could not move easily to difficult neighbouring keys on both sides.
This method gives a useful indication of the constraints of fingering patterns without resorting to
unwieldy complete listings that were feared when starting these analyses.
In the past, harmonics and difference tones have been invoked in many musical areas but do not
often lead to real explanations. For example, differences are the basis of tuning and harmony,
and it may appear that simple lists of lower pitches could account for the low frequency
response of the lute. There is little physical basis here, and such numerical exercises were not
even mentioned in Ref 21 on lute acoustics. In contrast, harmonics of strings will indeed drive
the various lute resonances, but since these are not known accurately a detailed matching of
numbers would not be helpful. However, in the case of resonances between strings, several
useful distinctions were made above, with a clear experimental basis. Even here, some expected
effects between adjacent strings were weak, while there were some strong resonances between
distant strings. This would be mediated by the lute itself, and may be worth future study.
Short aside on Climate Change
The harmonic series in maths is 1/1, 1/2, 1/3, 1/4 etc, or 1/n. The sum of this series does not
reach a limit for large n, and such ‘divergent series’ are not generally very useful. For interest,
the sum of 1/ns only converges if s is at least 2. The sum of 1/n2 is 2/6, for which the derivation
is not simple. For 1/n(n+1), and also for 1/2n, the sum is just 1.
Annual data of temperature, rainfall, etc have been kept for decades and centuries. This shows
record high and low values occurring over time.
For any randomly occurring events, the number of records r can be shown to be the sum of 1/n.
For example after 1 year n = 1, and there has been 1 record so r = 1. After 2 years the number of
records expected is 1½, because the second year can be either less or more than the first. After 3
years there are six possibilities, and in only 2 cases is the third year a record, so the number of
records after 3 years is 1 + 1/2 + 1/3 = 1.833. After 4 years 2.083 records are expected, then 2.93
records in 10 years, 5.19 records in 100years, 7.49 records in 1000years, 14.39 records in
1million years, etc.
In whatever way the meteorological data is examined, it is said that the expected number of
records is found, until about 1990. Then records begin to occur much more frequently, and
almost every year seems to produce some new extreme. This appears to be a proof by pure
maths that ‘something is up’, independent of detailed experiments, theoretical mechanisms,
modelling assumptions etc. The harmonic series can also be applied to traffic flow where each
bunch up of vehicles occurs after a record slow vehicle. The number of vehicles is n and the
number of bunches r. Other applications are testing components to a limit, collecting sets like
cigarette cards, and job interviews.
This topic was presented in a public maths lecture by John Barrow, and it was impossible to
resist the connection between lutes and current global concerns. Although playing any
instrument is very inefficient, in terms of sound energy output, lute players probably do not need
to limit their exertions, and in future may have to couple the right arm to a small generator.
Further evidence on Baroque lutes
Some additional support for the new ideas on baroque lutes can be found in Mersenne’s writings
in 1636 that ET was then the accepted tuning (eg Ref 9), referring most likely to his native
France. Also, before this date artworks rarely if ever show a definitive ET pattern of frets on
lutes. Ref 7 (p12) pointed to about three examples of ET and one of meantone, and many have
been happy to accept this view. However, from preliminary surveys in Ref 1, it seems that
virtually all depictions before about 1640 are exact or haphazard versions of equal spacings.
They occasionally recognize smaller spacings towards the body but not in sufficient detail to
identify a temperament. A meantone such as SCM or ECM has a complex fret pattern, but the
example suggested in Ref 7 (de Predis) is simply a jumble. At least two of the examples contain
problems of perspective, and an optical illusion where frets on the narrower part of a tapered
neck appear to have a greater spacing (Marziale). It would be unsound to believe a rare detailed
illusion arising from a common simple approximation. Rather than 3 early examples of ET to 1
of ‘meantone’, the survey in Ref 1 suggests the reversed proportions of perhaps 9 ‘equal
spacings and jumbles’ to 1 of ‘maybe something more definite’.
After around 1650, not to be taken too precisely for now, the very distinctive ET pattern is seen
regularly but not invariably. Examples by Vermeer were mentioned with the citterns, and a
further one is the famous portrait of Mouton, by de Troy in 1690. This unquestionably shows
ET, and with a reasonable bridge position, which is very impressive. The equally familiar
‘Company of Musicians and Singers’, by Puget in 1688, shows one of the earlier types of
confusion. There is a small spacing near the nut, as noted above for Vermeer’s lute, followed by
several equal spacings, and then some smaller equal ones as the artist begins to perceive a broad
trend. Interestingly, the ‘sharp’ types of meantone frets discussed above would resemble a
Pythagorean tuning, as discussed in Ref 1 and above. This may be worth bearing in mind when
assessing any reliable looking detail in the later paintings.
An alternative view is that painters took many years to recognize and accurately depict ET, but
such a position would require strong evidence. When studying Greek vases I noticed that
kitharai were invariably painted, for many hundred years, as though copied from a standard
template. There was less variation in their complex details than for the simple shell lyres and
aulos pipes, or even humans. Most classicists have no interest in such detail, but enthusiasts of
very early music could observe this in museums. It is likely that most kitharai differed from the
single image, occasionally used for modern versions.
Some further details of baroque frets are of interest. At a recital some years ago a theorbo player
announced that their next piece used ‘meantone’, then quickly moved several frets into slanted
positions. This can be explained with reference to a fret diagram and the discussions of
sensitivities. Frets 1, 3, 5, 6, 8, 10 need to be moved away from the nut. This can be done on the
treble side, which makes more use of fretted notes than the bass strings. Underlined frets need
moving most and some of the others might be unchanged. On the bass side where there is less
use of frets they might be moved slightly back to help a tight fit on the neck. This could even
reduce some stretch-sharpening. Frets 2, 4, 7, 9, 11 need the opposite shifts, which would be far
less satisfactory for the bass side. The result would obviously be an approximation, and only
possible for a small extra tempering above ET, both for tuning and stability. It should not be
imagined that the angles on the diagram are slanted frets, nor that meantone requires slanting. In
his lute tutor Andrea Damiani wrote that slanted frets might be used for 1/7th comma, with some
implied improvement. However, a good unison between the 1st and 6th courses on a renaissance
lute is likely to be a stronger constraint than for a theorbo making more use of open bass strings.
Setting good ET frets in D minor tuning may not always be straightforward. For example, in
ancient tunings fret 4 may have been flattened slightly. It has been suggested this would give a
better third between the 3rd and 4th courses (eg Ref 9, p41), but this would not be consistent with
most other fourths obtained between other frets. Early on I found fret 4 needed very slight
flattening even to set an exact ET, which also seemed inconsistent. This appeared to depend on
strings that are stopped at very different frets to produce important unisons with open strings.
The A strings use fret 5, whereas the D strings use fret 3. The detailed effects of even a slight
amount of stretch-sharpening could therefore be greater than in vieil ton which uses only frets 4
and 5 for the unisons.
Final remarks on Baroque lutes
All these suggestions may seem unusual to modern players who have unquestioningly started
with ET on a baroque lute, and also to the few pioneers who explore backwards to transitional
tunings. I did the same, and any thought of say ECM mixed up with several new tunings and
changes of strings would have seemed horrifying (Ref 22). However, the present inquiry
(‘istoria) attempts to recover the forward development by the original ancient geniuses. They did
not select styles from a collection on the museum shelf, which is the present ‘postmodern’
tendency. An initial ancient search for fine new sonorities and resonances would also have been
better served by tunings that were purer than ET.
Perhaps the only players to bring the transitional tunings to general notice with fine recordings
are Anthony Bailes and Sigrun Richter. It would be marvellous if a meantone, around say ECM
and SCM, could be used. All this music may be playable with such a temperament, and ET
would not be necessary. A greater consonance in these early baroque pieces might be very
satisfying, particularly for a player feeling an improved response of their lute. For D minor
tuning, in addition to the more likely historical use of ET, alternative enharmonic fret positions
could be tried, but extra lutes might be needed for large fret shifts of up to 8mm.
Now that French baroque lute music has become more popular, it is surprising that some
commentators are dwelling on its ‘refinement, precocity, superficiality’ etc. This sounds like the
old quip that France had decorators rather than composers. Fine performers of the above lute
composers and the harpsichord composers they inspired, and J S Bach himself, were drawn by
the expressive depth of the early French style. Some commentators even suggest a need for
German rigor and perfection, which only became a welcome rescue much later in the 18th
century. The bias is so engrained that some modern writings tell us that any outstanding early
pieces must have been influenced by Bach, even when the French publications predated his
birth. One also hears that emotional features of French music were caused by the repressive
court of Louis XIV, but the style was formed before his reign, and many of those responsible did
not have royal appointments.
Baroque Guitar and Bandora
A baroque guitar has the open string tuning of A d g b e. Uncertainties about octave strings
are not relevant here, and lute players seem less inclined to dispute this matter. The tuning is
spanned by 4 fifths, and it can be viewed as vieil ton for a lute without a top course a. The usual
comparison is a modern 6 course guitar with the major third interval between the 2nd and 3rd
courses. Throughout the 17th century the guitar developed into a fashionable instrument of high
art, with Corbetta teaching Louis XIV. The above comparison of tunings shows that the guitar
was using an open tuning equivalent to the lute’s vieil ton. Therefore, many pieces using a
similar range of keys can be expected to have been possible with a general meantone. Glancing
at some pieces reveals a suitable range, but specialists may like to make a more detailed survey
of pieces and the required enharmonics. Strangely, the new fashionable guitar essentially used
the old lute tuning, while the old aristocratic lute needed an early (r)evolutionary change to ET.
Perhaps the guitar followed suit, and several paintings such as Vermeer’s show definite patterns
of ET frets, which are generally used on modern copies.
We can return to wire strings, as used on a bandora. The basic six string tuning is usually given
(eg Ref 4) as a 7 course bass lute in D without the top string: C D G c e a (d), which spans
4 fifths, C to E. This is similar to the renaissance lute and a general meantone can be expected to
be useful over a similar range of keys. The top two courses would mainly have sharps, and the
effect of positioning the courses would be less severe than for the baroque lute.
Regarding a likely temperament for bandoras, the closely related orpharions have frets in SCM
or ECM. Also, an important use of the bandora was in duets with a cittern, which was near
SCM. As seen above, the cittern’s strings span 4 fifths, G to B, which might slightly restrict the
keys they had in common. The French cittern spans G to E, which lies within the bandora’s
range. These duets written for different instruments known to use SCM show that any small
complications in using a general meantone did not propel the ancients towards a simpler ET.
This specific conclusion has similar implications for all other types of instrument. The effects of
temperaments on consorts of lutes of different sizes and pitches were first analysed in Ref 2,
which may be useful here for consorts of citterns, bandoras etc. Readers who are familiar with
the repertoire may want to write out the enharmonics and choose the best alternative fret
positions.
There were also several types of large cittern where the tunings of the open strings may be
uncertain, and the present theory could help to indicate the likely open pitches. Bandoras were
also used for a continuo and may have needed to supply several enharmonic alternatives to
match other types of melody instrument. Modern makers often supply versions with extra frets
for certain notes and courses, but the lack of old specimens may not strictly confirm an original
use. There may be similarly sparse evidence for an original use of tastino lute frets. The only
sources seem to be Bermudo and Galilei, both of whom gave fretting schemes for ET. They also
had schemes for more general unequal semitones, which indicates that their ET was an
approximation for handling a wider range of keys, rather than a preference over a purer but
limited meantone. Any lack of minute tastini in paintings can scarcely be viewed as an
indication of a widespread ET.
Mandora and mandolin
For a final late addition, the family of mandoras and mandolins was encountered at a Lute
Society meeting. These are generally viewed as small lutes, but may have derived from early
gitterns. In the works of Chancy there are three short pieces for mandore, which I first
overlooked, but are of interest for the development of the later baroque mandolin. Ref 20 gives
the four open courses of the mandore as: c, g, c, f , which have intervals of fifths and fourths
like a gittern, but unlike a lute or guitar. The range can be viewed as the top string of a baroque
lute to a tone below the 5th course, but an octave higher. The tiny mandore therefore required a
string length about half that of a baroque lute, similar to the 35cm of the English cittern (or
gittern) analysed above. Examination of the tablature pieces shows that a general meantone
would fit every note. All courses use the naturals on frets 0, 2, 4, 5; the 1st course needs fret 7;
and the 2nd course needs fret 1 for C# and fret 3 for Eb. This fret 3 can also provide Bb on the 3rd
course. It is seen that the open strings are spanned by only 2 fifths, which provides more
common enharmonics than vieil ton or a cittern. The disadvantage is the wide intervals between
strings. This would inhibit fingering patterns, and reduce the number of notes available for the
more complex music of a lute. The lute’s early avalée experiments were similar, and the final D
minor tuning was a surprising return, but with the difference of the F natural courses.
The surviving examples of the baroque mandolin seem to be later 18th century types with fixed
metal frets in ET, and they probably acquired metal strings. The loud high pitch was used for
many concertos, and ensured survival as a popular portable instrument, unlike the lute. The main
interest here is the open tuning of the six courses as: g b e a d g. Once gain, this seems
close to the lute’s vieil ton, but with the 4th and 5th courses a semitone lower, which will seem
odd to lute players. The mandolin’s open strings are spanned by 4 fifths, G to B, and may
initially suggest a general meantone. Flat enharmonics on frets 1 and 3 would be fine for courses
4 and 5, but one can see that the standard fret 2 fixes C# and F#, and fret 4 fixes D# and G#, while
the G courses have many flats. The baroque mandolin would have needed ET, similarly to the D
minor lute tuning. As a strong contrast, at the same meeting I learnt (Ref 6) that a bandora for
continuo use had just been made to special order in QCM, with extra frets 1, 6, 8 on the top
course(?).
Finally, an astonishing consequence emerges from this new analysis of open string pitches and
the use of a general meantone. If ET had been a long standing system then any open tunings
might have been used over the centuries. For the likely situation in which ET later became
widespread, all sorts of new open strings would have then become permissible, but did not
appear. Even now one could devise a new modern guitar with some very unusual open tunings.
A random choice such as Eb Ab d f# b f  spans 10 fifths and would be virtually unusable in a
general meantone, but perfectly tuneable in ET. The ancient constraints were musical
foundations on intervals such as the fifth; availability of notes on neighbouring strings; technical
improvements for adding lower and higher strings; the ability to tune meantones other than ET;
and the production of resonances. Over history many of these factors diminished but the ancient
choices of open tuning have been retained, largely by convention.
USING a C40 METER for ANY TEMPERAMENT
This section presents a method for tuning all these enticing meantones using only a basic
electronic meter. Many players who use a meantone and a meter are probably restricted to SCM
because it can be approximated well on deluxe tuning meters, which may have only a few preset
temperaments. Valotti’s temperament for keyboards, and especially organs, may be the only
preset useful for lutes. Ref 2 outlined how the theory of any temperament could be used to
calculate deviations from the ET system on all basic tuning meters. I recently bought a C40 for
£10, after struggling ‘off and on’ with a forty year old basic but expensive meter, which was
also sluggish and insensitive. It seemed likely that a new deluxe model would not have
sufficient extras, and might even have restricted flexibility.
The Korg C40 with its little pickup works well for ET, but one notices that it operates best near
the central null reading. As one tunes a lute the needle seems drawn to this null point, like an
encouraging ‘pat on the back’, but it becomes uncertain and erratic for much of the 100cents on
each side. The smallest division is 5cents but one can estimate 1cent, which is more precise than
generally obtainable or required on a lute. There are marks for perfect major and minor thirds, at
–13.7 and +15.6cents, but these are not of general use and are hampered by the effects away
from the null point. The only extra feature is changing an A of 440Hz in simple steps of 1Hz
down to 410 and up to 480Hz. This looked potentially useful, and some trial calculations have
developed into a general method for any temperament on any instrument.
SCM is a convenient and useful start. All the calculations need to use the reference of ET. This
requires an intermediate step from the perfect Pythagorean intervals normally used as the
reference in defining temperaments. It can be seen from Ref 2 that SCM is defined by a fourth
that is 1.6cents wider than ET, which in turn is 2cents wider than perfect or PYT. A general
derivation is given later.
The first step is to explore how the interval of 1.6cents compares with a step of 1Hz from
440Hz. The interval is 1+ 0.0595(1.6/100) = 1.00095. This produces gives a pitch of 440.418Hz,
which is almost ½Hz higher than 440. Further intervals of 1.6cents give successively:
440.836, 441.255, 441.674, 442.094, 442.514, 442.934, 443.355, 443.776, 444.198. The lower
fourths are 439.582, 439.165, 438.748, 438.332, 437.916. This indicates a method of tuning
successive fourths above and below an A at 440Hz, by altering the meter setting between 438
and 444Hz, together with very small or zero departures from the null point.
To tune a D, one sets the meter at 440 and estimates 1½cents above the null reading.
Then for a G one can reset to 441 minus about ½cent, or stay at 440 plus about 3cents.
For C one can stay at 441 plus 1cent, or try 440 plus 5cents.
The required deviations can be estimated by noting that the increments of about 0.4Hz are about
1½cents (or c). This is accurate enough and saves unnecessary detailed calculations, but
accumulating rounding errors need attention. The procedure continues with:
F - move up to 442 minus 1½c, or stay at 441 plus 2½c, or 440 plus 7c.
Bb - 442 at the null point, or 441 plus 4c, or 440 plus 8c.
Eb - 443 minus 1½c, 442 plus 1½c, 441 plus 5½c, or 440 plus 10c.
Ab - 443 at the null point, 442 plus 3c, 441 plus 7c, or 440 plus 11c.
Db - 443 plus 1½c, 442 plus 4½c, 441 plus 8½c, or 440 plus 13c.
Gb - 444 minus 1c, 443 plus 3c, 442 plus 6c, 441 plus 10c, or 440 plus 14½c.
Cb - 444 plus 1c, 443 plus 4½c, 442 plus 8c, 441 plus 11c, or 440 plus 16c.
The fourths below A follow similarly:
E - move down to 439 plus 2c, or stay at 440 minus 1½c.
B - 439 plus ½c, or 440 minus 3c.
F# - 439 minus 1c, or 440 minus 5c.
C# - 438 plus 1½c, 439 minus 2½c, or 440 minus 6c.
G# - 438 at the null point, 439 minus 4c, or 440 minus 8c.
This method can be employed for every note on the fingerboard, whereas a deluxe meter may
miss many notes. All the alternative settings could be tried, or just those nearest the null point.
In practice a sequence of nudges either side of null would be most satisfactory. The best way to
use the method would be a neat handy list of preferred settings. Open strings and fretted notes
can be tuned and cross-checked in any preferred order. The more remote fourths that may be
unavailable with a deluxe meter have larger departures from ET, and can therefore be the most
valuable for overall tuning. For example, Valotti tempered only six fourths by 1/6th of a
Pythagorean comma, which has a negligible difference from the syntonic comma. The other
fourths were left perfect and are less useful for a lute. This produces a return to the tonic, unlike
all meantones, except ET. Four further meantones, ECM or 1/8th comma, QCM or ¼ comma,
1/7th and 1/5th comma can be listed, and may help players explore away from SCM or ET.
First it is useful to show how definitions relative to perfect intervals can be converted into
deviations from ET. From Ref 2 the fourth in ET is the solution of 12 = 32, which is 1.3348.
A perfect fourth is 4/3 = 1.3333, so ET is sharper by 1.001125. This can be expressed in ne cents
as 1 + 0.0595ne /100, where ne = 11.25/5.95 = 1.9c, which is often taken as 2c.
Next, an ‘x comma’ meantone has a fourth (4/3)(81/80)x  (4/3)(1+ 0.0125x). This is wider than
perfect by n = 12.5x/5.95 = 21.0x cents. For SCM, n = 21/6 = 3.5c, so the fourth is wider than
ET by 1.6c, as used above. This procedure can be applied to any meantone, and any other
temperament expressed relative to the perfect fourth or fifth. The above relations can also show
that ET is close to 1/11th comma, since x = 1.125/12.5 = 1/11.1.
ECM has a fourth that is only 0.7cents wider than in ET, producing an interval of 1.00042. This
is less than half that for SCM, because the ET reference is already about 1/11th comma and not
pure PYT. The only opportunity or need for changing the meter setting occurs between 442 and
439Hz. A series of ten fourths from A to Cb is only 7cents wider than ET, and it might be
possible to estimate all notes with the meter set at 440Hz. The preferable alternatives close to
the null point are:
A - 440 at null point. Db - 441 plus 2c, or 440 plus 5½c.
D - 440 plus about ½c. Gb - 442 minus 1½c, or 440 plus 6½c.
G - 440 plus 1½c. Cb - 442 minus ½c, or 440 plus 7c.
C - 441 minus about 2c, or 440 plus 2c. E - 440 minus ½c.
F - 441 minus 1cent, or 440 plus 3c. B - 440 minus 1½c.
Bb - 441 at null point, or 440 plus 3½c. F# - 439 plus 1½c, or 440 minus 2c.
Eb - 441 plus ½c, or 440 plus 4c. C# - 439 plus 1c, or 440 minus 3c.
Ab - 441 plus 1c, or 440 plus 5c. G# - 439 plus ½c, or 440 minus 3½c.
QCM, which is theoretically significant but practically inconvenient, has fourths that are
3.3cents wider than ET, with an interval of 1.00196, just over twice that for SCM. This gives a
pitch of 440.862 relative to 440Hz, so that the setting can be changed with almost every
successive fourth. The following list gives the setting close to a null point, and also the basic
setting of 440Hz that is practically almost impossible. Intermediate settings could be deduced by
a reader, using the above procedure.
A - 440 at null point. Db - 447 at null point, or 440 plus 26½c.
D - 441 minus about ½c, or 440 plus 3½c. Gb - 448 minus ½c, or 440 plus 30c.
G - 442 minus 1c, or 440 plus 6½c. Cb - 449 minus 1c, or 440 plus 33c.
C - 443 minus 1½c, or 440 plus 10c. E - 439 plus ½c, or 440 minus 3½c.
F - 443 plus 1½c, or 440 plus 13c. B - 438 plus 1c, or 440 minus 6½c.
b
B - 444 plus 1c, or 440 plus 16½c. F# - 437 plus 1½c, or 440 minus 10c.
Eb - 445 plus 1c, or 440 plus 20c. C# - 437 minus 1½c, or 440 minus 13c.
b
A - 446 plus ½c, or 440 plus 23c. G# - 436 minus 1c, or 440 minus 16½c.
1/7th comma has fourths that are 1.1cent wider than ET, which is an interval of 1.00065, or a
pitch of 440.286Hz. A span of 15 fourths can use intermediate settings from 443 to 439Hz.
A - 440 at null point. Db - 442 plus 1c, or 440 plus 9c.
D - 440 plus 1c. Gb - 443 minus 1½c, or 440 plus 10c.
G - 441 minus 1½c, or 440 plus 2c. Cb - 443 minus ½c, or 440 plus 11c.
C - 441 minus ½c, or 440 plus 3½c. E - 440 minus 1c.
F - 441 plus ½c, or 440 plus 4½c. B - 439 plus 1½c, or 440 minus 2c.
Bb - 441 plus 1½c, or 440 plus 5½c. F# - 439 plus ½c, or 440 minus 3½c.
Eb - 442 minus 1c, or 440 plus 6½c. C# - 439 minus ½c, or 440 minus 4½c.
Ab - 442 at null point, or 440 plus 7½c. G# - 439 minus 1½c, or 440 minus 6½c.
1/5th comma is a practical extreme that has fourths 2.3cents wider than ET, which is an interval
of 1.00137, or a pitch of 440.603Hz. The settings can range from 446 to 437Hz, entailing a few
more settings than SCM.
A - 440 at null point. Db - 445 minus 1c, or 440 plus 18½c.
D - 441 minus 1½c, or 440 plus 2½c. Gb - 445 plus 1½c, or 440 plus 21c.
G - 441 plus 1c, or 440 plus 4½c. Cb - 446 at null point, or 440 plus 23c.
C - 442 minus 1c, or 440 plus 7c. E - 439 plus 1½c, or 440 minus 2½c.
F - 442 plus 1½c, or 440 plus 9c. B - 439 minus 1c, or 440 minus 4½c.
Bb - 443 at null point, or 440 plus 11½c. F# - 438 plus 1c, or 440 minus 7c.
Eb - 444 minus 1½c, or 440 plus 14c. C# - 438 minus 1½c, or 440 minus 9c.
Ab - 444 plus 1c, or 440 plus 16c. G# - 437 at null point, or 440 minus 11½c.
For all these meantones, fret positions can be found from Ref 2, or interpolated with sufficient
accuracy from the diagram. However, each meantone has a different list of meter settings, which
requires separate calculations that cannot be inferred from another list.
The alternative settings, in addition to tuning one’s lute, can also check the calculations, and
even test the internal consistency and accuracy of the meter itself. This latter test could be made
irrespective of any meantone, so please tell Korg and me if you find a problem.
The method can also be adapted to different pitch standards such as an A at 415Hz. For
example, in SCM the interval of 1.00095 for 1.6cents gives the pitches: 415.394, 415.789,
416.184, 416.579, 416.975, 417.371, 417.768, 418.165, 418.562, 418.959. Comparing this with
the previous list, the increments of 1Hz are largely correct, but the decimals decrease by about
5%, or 25/440. This means that the fine nudges about the null points ideally need recalculation,
but may remain satisfactory for practical tuning. For example, even after 10 fourths, Cb would
use 419HZ at the null point, whereas the previous setting was 444 plus ¾c, only about 1c more.
A related possible problem is the lower limit of 410Hz for an A on the meter. This could be
avoided by labelling A a semitone, tone etc lower. Any practical temperament for any kind of
instrument should be possible, with a clear head and a neat list. The meter is just a continuous
scale of cents marked off in 1200c for an octave, 100 for a semitone, 5 for a division, and 1
estimated. The aim is to use the labelled reference points of A, Eb etc as conveniently as
possible. For example, if a certain temperament had an Eb that was say 9cents sharper than ET,
one would want to set A at 9c above 440Hz, which is 442 plus 1½c, similar to F in 1/5th comma,
Eb in SCM, Db in 1/7th comma etc in the above lists. Most keyboard systems will be milder than
the remote notes in strong meantones.

ORPHARION by PALMER
This orpharion initially appears to be amenable to a precise analysis of the temperament, and
also some further comparisons between the courses of different lengths operating with the
slanted frets. However, it will be seen that there are some considerable discrepancies, which can
be shown to be errors in marking out or making. A highly detailed treatment, as first envisaged,
is not justified or necessary, and useful conclusions can be more general.
The measured fret positions on the drawing by Darryl Martin (Ref 5) were checked against the
drawing and a photo, and can be listed as:
Treble side
Nut 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Bridge

mm 30.5 55.5 83.3 104.5 129.9 151.3 173.7 192.7 210.9 228.9 245.3 261.0 277.3 290.3 304.1 518.5

Bass side
Nut 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Bridge

mm 35.8 64.5 97.0 120.1 149.8 174.9 198.7 221.7 242.4 263.6 282.5 301.4 - - - 601.5
The bass side can be compared directly with Rose’s positions:
Nut 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Bridge

35.0 64.2 94.0 122.2 150.2 175.0 199.1 222.3 242.6 261.5 280.7 298.7 315.8 331.9 346.0 599.1
The virtually identical string lengths close to 600mm may suggest some old standard, close to a
modern one. The similarities and differences from Rose need to be explored in the analysis.
Frets 3, 4, 10, 11, 12 differ most. Palmer’s fret 12 is nearer to the bridge than the nut, unlike
Rose’s. This indicates that the saddle position on the bass side should be at least 602.8mm, or
about 604.5mm in comparison with Rose and consistent with some stretch-sharpening. This is
3mm nearer the tail, and the belly has signs of bridge alterations. On the treble side the
corresponding positions need to be 522 and 523.5mm, which is 5mm nearer the tail. Another
aspect of the slanted frets is that the measurements were made at the end of the frets, not under
the strings, but scaling will largely take account of this. Uncertainties are possible in the nut
position or an equivalent fret. Also, in Ref 5 there is a question over earlier different
measurements for the treble side, but the photo suggests these later positions are better.
Considering these factors, the fret analysis initially takes fret 12 as a reference, and the fret
positions are plotted on a diagram of temperaments. Frets 1 to 12 are shown by circles for the
bass side, and crosses for the treble side. The mean scaled values are shown by the combined
circles with crosses, for easy comparison with Rose’s simple perpendicular frets. Frets 13 to 15
on the treble side are shown by squares, but no perfect fifths on lower frets are given, in view of
a problem with fret 7. The solid dots represent a shift of the nut 1.5mm away from the mean
position of the frets; and the dots in brackets a reduced shift of 0.75mm. The crosses in brackets
indicate a change in the position of fret 12, as discussed later.
The overall impression is that there is less scatter than in the theoretical schemes in Ref 1, but
more than in Rose’s frets. For comparison with Rose and further analysis it is useful to list the
nearest systems for the averaged fret positions as:
Fret 1 2 3 4 5 6 7 8 9 10 11 12
ECM SCM ECM QCM (PYT) PYT (QCM) PYT SCM PYT ECM
/SCM /QCM /ET /ET /QCM
Fret 13 14 15
SCM PYT QCM
/QCM
All the frets, except 5 and 7, lie between PYT and QCM, whereas Rose’s reached only SCM.
Broadly, the important frets 1 to 4, and also 9, 11,13, 15 are ECM to QCM, but frets 6, 8, 10 are
ET to PYT. The frets with a similar tendency to Rose’s are 1, 2, 6, 10, 11, 13, 14. The best
overall description is about 1/7th comma meantone. This might be considered remarkable for the
date of 1617. However, this fact is consistent with the analyses above, and Ref 1 on lutes, even
though it has often been assumed that ET was standard by this time. The 1/7th comma is only
slightly different from an SCM generally found from previous matches against standard strips,
and often used for modern versions. It is only slightly more tempered than Rose’s 1/8th comma.
Previous estimates of SCM for Palmer and near ET for Rose were slightly further apart.
However, frets 5 and 7 have greater deviations from these systems. Some special examination is
needed, but finally this does not greatly alter the above conclusions. A comparison of the bass
and treble sides is of interest, and this will also help to simplify further analysis. For values
scaled to 600mm, the differences ‘bass minus treble’ can be listed as:
Fret 1 2 3 4 5 6 7 8 9 10 11
mm 0.6 0.4 0.8 -0.6 -0.2 0.2 -1.9 -0.8 -1.1 -0.7 -0.7
There is moderate consistency between the bass and treble sides for most frets. This means that
detailed analysis of most frets can use a single average value. The largest problem occurs with
fret 7. The scaled treble position is 199.6mm close to ET, but the bass position is 197.7mm. This
is way off the diagram, past TCM and close to ½ comma at 197.4mm. A whole comma flatness
would be an exact 195mm (13/40 x 600). The bass side is 1.0mm flatter than QCM and 1.8mm
flatter than ET. This indicates a significant error in marking out and/or making. A slightly
skewed position of fret 7 can be discerned on the photo, but is not immediately obvious in
relation to the varying degrees of slant for the other frets. The error is unlikely to have arisen
from any initial theory, construction or adjusted tuning, since the treble side is good. However,
the effects would be less than might be expected because there are 9 courses, and the lower three
or four courses would not use fret 7. This roughly halves the error, and the flattest bass position
in use is about 198.6mm, which is in QCM.
Fret 9 has the second largest difference between the bass at QCM and treble at SCM, but the
deviation from an average is a moderate 0.5mm. The other frets show smaller differences, and
no use of fretted notes on the lowest courses would reduce the effective deviations. In the above
list, the differences begin as positive on low frets and tend to negative for high frets. Since the
central frets 5 and 7 are almost perpendicular to the strings, it is seen that both low and high
frets are slightly more perpendicular to the strings than required. This could have resulted from a
slight bias in marking up or sawing at the bench.
Fret 5 lies way beyond PYT. For the bass and treble sides the scaled positions are 149.1 and
149.3mm. These are mutually consistent, rather than skewed as for fret 7, but about 0.8mm less
than PYT and 1.7mm less than SCM. This would be a considerable error in tuning since 1½mm
uncertainty in a fret spacing of about 25mm is 6cents. It is difficult to determine whether there is
a relation to the discrepancy on fret 7, or stretch-sharpening effects are involved, even after
considering the several lengthy transformations discussed above.
A useful observation is that all of fret 5 and the bass of fret 7 are flatter than SCM by 1½ mm, so
that their relative positions are reasonable. Adding this 1.5mm at the nut for the bass side of all
frets produces the solid dots on the diagram. Frets 5 and 7 are clearly moved to between ET and
SCM. Fret 2, with a tempering slightly less sensitive to position, is moved in the opposite
direction to ET. Overall, the lower frets become more scattered, while the higher frets are
slightly less scattered, but the average temperament is scarcely changed. The same process was
applied to the frets of Rose and Campi, and is plotted on their diagrams. The effects and
conclusions are similar to this treatment of Palmer’s instrument.
As explained for Rose’s frets and the citterns, measuring real frets 2, 5, 7, 10 is not a sound way
of identifying a tempered system. The variations of position could be due mainly to errors in
setting up, or string elasticity. Even if these details were known precisely, a deviation of say
1mm in fret 7 would be a tuning error in any tempered system, and of no use in classifying a
particular system. In contrast, differences of say 3mm in frets 1 and 3 are very useful indicators
of temperament. This caution may be helpful for a recent label of ¼ comma for some of
Palmer’s frets that are in fact way beyond QCM (Ref 5). This probably came from the last
comparison strip in a series. More seriously, meantones are sometimes labelled theoretically by
the value of the fifth in cents, as in Ref 8. If this were taken literally, from a single measurement
on a fret 7, the conclusions are likely to be meaningless.
Next, the patterns of spacings between frets can be compared with Rose and the citterns. This is
a further way of highlighting how Palmer’s frets deviate from SCM and other systems.
For a theoretical SCM scaled to 600mm the pattern is:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12 12/13 13/14 14/15
36 28 | 32 25 | 29 27 | 21 | 24 19 | 22 17 | 19 || 18 14 | 16
| | || || | | || |
The sequence of long and short spacings is:
| L S | L S | L | L S | L S | L S | L || L S | L
For Palmer’s frets the scaled mean spacings are:
0/1 1/2 2/3 3/4 4/5 5/6 6/7 7/8 8/9 9/10 10/11 11/12 12/13 13/14 14/15
35 29 | 32 23½ | 29 25 | 25 | 22 21 | 21 19 | 18 || 19 15 | 16
The first three spacings are good, then 3/4 is small because fret 4 is close to QCM. The spacing
4/5 looks good, but results from the flat fret 5 together with the small 3/4. The most problematic
spacings are underlined. Fret 6 is the ancient customary midpoint, but flatter like frets 5 and 7.
This also enlarges the smaller spacing 6/7 of an exact SCM. The characteristic alternate large
and small spacings are lost for all the higher frets, although there is a slight recovery for the
second octave. The higher frets have a pattern of a disordered ET, or ECM with the equal 8/9
and 9/10, similar to the citterns and Rose. This appeared to arise from the ancient difficulty of
constructing higher positions. The old methods may have been perfect intervals, fifths based on
lower frets as used by Dowland, simple linear divisions as used by Ganassi, and simply filling
the space between frets 7 and 12, all subject to inaccuracies in making. In later times, complete
characteristic ET patterns may have developed in this way, as much as by any theory. However,
in the present cases there is no indication of any specific method for the higher frets, such as
Bermudo’s or the 18/17 rule, and clearly no trace of ET tuning in the important lower frets.
The individual frets need mentioning only briefly, since many features and probable derivations
have already been outlined. (With each further instrument the comparisons become more
intricate, and there may be some changes as the writing has proceeded.)
Fret 1 is SCM or 1/7th comma, between Rose’s ECM and the citterns. The position could have
derived from Gerle’s scheme of 80 years earlier, the perfect interval 16/15, the citterns, Rose, or
Dowland writing only 7 years earlier.
Fret 2 is also SCM and, as for the other instruments, most likely needed a fine adjustment of a
perfect interval construction.
Fret 3 is about ECM, which is close to the SCM of citterns, and appears much better than the
unusual position near PYT obtained by Rose and Dowland. The origin could have been Gerle’s
scheme, or an average of perfect intervals, and fine adjustment.
Fret 4 is QCM like the citterns and probably derived from the perfect interval. Rose’s better
position in ECM needed fine adjustment.
Frets 5 and 7 have been examined above, but their problems cast a shadow over higher frets. It
is notable that both Rose’s and Palmer’s higher frets have the gradation of ET, as also seen in
the citterns. The different temperings for frets 8 and 9 may be related to differences in fret 7.
Fret 6 is near the midpoint of 5 and 7, but closer to PYT than the Rose’s ET or the SCM for
citterns.
Fret 8 is beyond ET, and far from Rose’s SCM. It is similar to the English cittern but not related
to any fifth on fret 1.
Fret 9, conversely, is beyond SCM, and far from Rose’s ET, but similar to citterns and could be
a perfect fifth from fret 2.
Fret 10 is close to PYT like Rose’s, but unlike some citterns. It might derive from a perfect
interval, but is not a perfect fifth on fret 3.
Fret 11 at ECM is also quite close to Rose’s and some citterns. It could be a perfect interval, but
not a perfect fifth from fret 4.
Fret 12 appears satisfactory as a reference. Fret 13 is slightly more tempered than fret 1. Fret 14
is also slightly sharper than fret 2, and this places it beyond PYT. It was probably an unadjusted
perfect interval as discussed above for several other cases. Fret 15 in QCM is sharper than fret 3.
With Rose’s simple perpendicular frets, the higher octave frets 13 to 15 were found to be an
almost half-size repeat of frets 1 to 3. This provided a check on fret 12 and the overall
consistency. For Palmer’s treble side, fret 13 gives 277.3 – 261.0 = 16.3mm, compared with
15.3 or half 30.5. Fret 14 has 290.3 – 261.0 = 29.3, also larger than 27.8. Fret 15 has 43.1,
compared with 41.7. This indicates that both octaves might share the same temperament, or be
in tune, if fret 12 were moved about 1.3mm further from the nut, leading to the relative upper
positions 15.0, 28.0, 41.8mm. However, moving a single fret would be completely arbitrary, and
a precedent for tinkering with other frets. This particular shift would also force frets 5 and 7 into
even flatter positions, but not greatly affect the temperaments deduced for other frets. For fret 1
the frequency ratio would be reduced from 1.0620 to 1.0618, which is about 0.3cent, while for
fret 11 the flattening is a larger 4cents. The resulting change of temperament for the treble side
of each fret is shown on the diagram by crosses in brackets. As expected, the difference is
greater for the higher frets. The scatter increases, which is not an improvement, but the overall
picture is unaltered. More generally, it is also seen at each fret that the various points for about
six conditions form a moderate cluster. This gives some confidence in the conclusions.
Summary of methods
It may be helpful to list the various types of transformation involved in analysing the real frets.
In order to be useful these should represent some physical effect that can be applied universally
to all frets, rather than changing a few odd frets for a better fit. Ideally, transformations would
place most frets in a single consistent temperament on the diagram. It would be almost as useful
if some average temperament were insensitive to a particular change, and this was the most
frequent result.
i. Addition at the nut, which increases the distance of each fret from the nut. This accounts for
the fact that an open string has no stretch-sharpening, and allocates a simple equal amount at
each fret. It affects the lower frets most, and was used in the above analyses.
ii. Addition at each fret proportional to the distance from the nut. This accounts approximately
for stretch-sharpening due to an action height that increases towards the bridge. The new frets
scale to the initial positions so that no special analysis was needed.
iii. A simple combination such that the shift of each fret increases as in (ii), but from the non-
zero value in (i). The scaled result is slightly less than for (i). A shift that decreased from the nut
would have relatively more effect for lower frets.
iv. A special calculation for each fret could use the elastic relations derived above. This would
require an accurate modulus, and also fine detail on the effective heights of the nut, bridge and
each fret. It would be better to use the full theory to avoid mistakes, but simplifications may be
possible for special cases. For making some general deductions about the extra stretch x, a
compact form that might be used with care is m2 L /(L - l1 )l + c( L - l ). The first term for
pressing between frets would roughly double from the nut to fret 12, and treble by fret 19. The
generally smaller second term for the action height would increase linearly from the nut.
v. For making an instrument a full calculation could be a clearer method than juggling many
separate semi-quantitative trends. Analysing an existing instrument is less straightforward and
juggling has also been necessary. A combined approach can also lead to new useful relations.
For example, a 2mm subtraction at the nut during making, which is equivalent to an addition for
analysis, initially seems to assign a simple equal degree of stretch-sharpening to all frets.
However, in the theory it was seen that an equal depth of pressing gives more sharpening on
higher frets. The simple subtraction is therefore equivalent to a lighter pressing on higher frets.
This may be a natural tendency with the smaller spacings, especially for more rapid passages. A
simple method therefore contains a subtle effect, and its use in making a cittern could encourage
or dictate the player’s touch. As a quantitative measure, for frets near 12 the spacings are about
half those near the nut, so the effective depth can be reduced by a factor of about 0.7 (or 1/2) to
produce the same amount of sharpening.
vi. Alternate large and small spacings, as in SCM, were found to produce smaller and larger
stretches. The result on temperaments could be explored on the diagram by increasing the
1.5mm addition on the more sensitive frets 2, 4, 7, 9, 11. More simply, the effect is illustrated by
halving it on the other frets 1, 3, 5, as shown by dots in brackets. There is little difference, but a
flattening of frets 1 and 3 reduces 0/1 and increases 1/2, then reduces 2/3 and increases 3/4. This
would change Palmer’s scaled spacings to 34, 30, 31, 24½mm, which is nearer to an ECM with
35, 30, 31, 27mm. Stretch acting on alternating spacings produces a slightly less tempered
system. As early as the initial Ref 2 a tendency towards ET was anticipated, while looking out
for possible physical mechanisms on lutes. Stretching on ET frets would produce a sharper ET
pattern, as in (ii) above, and this may apply to the higher random frets. A system nearer PYT
would be made more tempered by stretch, indicating some convergence towards ET.
vii. The effects in (vi) will also combine with variations of pressing for different frets, as in (v),
and also variations between courses. This complexity would benefit from the calculational
approach in (iv).
viii. The transformations all produce greater relative effect on lower frets. This arises because
the changes also revise the position of a bridge, which is often practically sensible and
acceptable for these instruments. The only way to influence the higher frets would be some
deviation that increased very strongly towards fret 12, but there is no physical basis for this. To
influence the status of higher frets, even slightly, required an abrupt arbitrary change as seen
above in analysing Palmer’s fret 12. A related point occurs in classifying the tempering of the
higher frets below 12. This requires an extremely accurate scaling factor. The effect became
particularly noticeable for the small additions in (i), which should be consistent on all the
diagrams. This subtlety is not inconsistent with the visually random patterns of higher frets.
ix. While the fret measurements on both orpharions and one cittern had a nominal accuracy of
0.1mm, for the other citterns the accuracy was plus or minus ½mm or even 1mm. Large random
errors could bracket ideal tempered positions more easily than the single direction of a
transformation. However, it is clearly not legitimate to think that poor measurements can
produce a more significant match.
x. It may be imagined that a few frets could be positioned with less trouble, or a suitable
scheme arise purely by chance over the early decades - or even the length of this writing. The
low probability of this can be seen as follows. If there were say 3 possible positions for each
fret, perhaps including 1 mistake, then the chance of setting up only 4 specific important frets
would be 1 in 81, which is bad enough. For a more realistic 4 positions on each of 12 particular
frets the chance is 1 in 250 million.
Final comparisons of tuning, and survival of instruments
The detailed analyses lead to the firm conclusion that the various wire-strung instruments have
meantone tunings. The orpharion of Rose is 1/8th comma, Palmer’s is 1/7th comma, and the
citterns are mainly 1/6th comma. However, there are many departures of differing degree on
many frets. On all instruments the higher frets 8 to 11 resemble a disorganized equal
temperament, but the second octave is a reasonable match to the first. The citterns often have a
sharper fret 1, almost in ¼ comma. This is consistent with the music requiring only flatter notes
F, Eb, Ab, C, rather than the two sharps needed on a lute. Importantly, the sharp fret 1 would
have some further stretch-sharpening, but this would be a smaller amount and more controllable
than for an initially flatter fret with a smaller spacing from the nut. This could assist almost
perfect chords on the lower frets.
The analysis of Rose’s frets was consistent with finely tuned low heights for the frets and action,
such that any universal types of correction would degrade the tuning. Palmer’s frets were more
scattered or in error, but universal corrections could not give much better tuning, and individual
frets would need correcting. The cittern frets were similar to Palmer’s, but would probably have
greater scope for useful stretch-sharpening.
Many further details of temperament and tuning were analysed, and two complementary
viewpoints are needed for a sensible judgement. It is legitimate to assess in modern terms the
accuracy of the surviving tunings, but we also need to take into account the probable way the
ancients worked. This can help us to understand their original results in a better context than
familiar straightforward modern precision.
An ancient maker’s strip with fret positions would have derived from initial constructions and
ideas. Then, most likely, adjusted positions were obtained by sounding actual strings on lutes
and wire-strung instruments. Lute frets could be adjusted later, whereas strips for making more
than just a single wire-strung instrument would need good adjusted positions. The player might
also need a strip for a lute, but not for a cittern. In addition, the maker’s experience and feedback
from playing customers might have nudged some positions. Changes and errors could occur
while making the strip, as appeared to affect Dowland’s scheme. These strips could also be
copied, scaled for different string lengths, and adapted for different instruments. The length
scaling effects and angling required for an orpharion and bandora would be intricate,
particularly in setting up a new design. The final strips could be an amalgamation of many
factors, including unusual features and errors, even before starting to mark out a fingerboard.
These same templates would have been used repeatedly. It is unlikely that strips would be
‘recalibrated’ often, and some checking might be no better than a simple ruler perhaps with
large 1/12th inch (2mm) divisions, or an irate purchaser.
Deviations in cutting and fitting could result from an indistinct line rather than use of a marking
knife, or cutting the wrong side of a line. The positions at a bench can produce a bias, and
merely turning the work round or working from another side are simple cases. The slightly
incorrect fan of angles formed by Palmer’s frets is a likely example. Unfinished fitting of
wedges might have left us with frets in unintended positions.
For ancient craftsmen, ‘laying down a line’ in a precise position with no simple relation to other
points would be unusual and quite difficult. Most ancient work proceeds from a series of
references. For example, the type of accuracy in finishing the body of a lute, fine furniture, or
wood-carving is not that found in modern engineering or computer-aided design. Conversely,
apparently simple ancient forms are often enlivened by subtle deviations – the Parthenon is the
famous case. Accurately placing a fixed fret is a process more suited to a modern maker, using a
calculated position, a precision ruler and a tuning meter, working more slowly and cautiously in
a technological manner. Modern copies are generally made to special order in fewer numbers,
and poor quality might be less acceptable. Larger scale production of early types of instrument
has often been poor compared with modern types, and some of this variability can indicate what
might occur in ancient workshops.
The tuning accuracy that the ancients achieved with the fixed frets was found above to range
from better than ½ mm, commonly 1mm, and occasionally 2mm or worse. In some cases
difficult frets were accurate, as with fret 1 and the fine tempering of fret 2. Sometimes simpler
positions were poor, as with frets 5 and 7. The analysis made a clear distinction between frets
that determined an intended temperament, and other frets that would simply be out of tune in
any system. In more detail, apparently flat frets could be sharpened by a player, whereas a sharp
fret would be a greater problem. Frets might not have had their scallops, fret wire or wedges
adjusted by a maker, but there would be a limit to individual adjustments in order to retain an
even feel from fret to fret. Surviving ancient wedges are almost always behind a fret, which
suggests they were rarely if ever used for flattening. These wedges would be close to a lightly
pressing finger. Fine adjustments of frets and wedges might be reversible, and easily renewed
when worn down. This could reduce damage to the main fingerboard. Wedges may seem
mysterious, but perhaps their obvious use as easily fitted strong packing, and some colour
coding, are sufficient purposes.
The present study has used only a little experiment, but a full program on ancient frets could use
a test rig. An ultimate analysis would require accurate knowledge of all the fret heights etc, and
it would be useful to know how much an instrument had been played. After all this detail,
distortions over the centuries might have altered the sensitive action. Considerable effort and
dedication would be needed to set up exact tests of old frets, especially when they might turn out
to be poor and of little commercial use.
It is useful to consider the three levels of accuracy or error in relation to practical methods, both
ancient and modern. Interestingly, these are all within an ancient 1/12th inch division, compared
with a modern accuracy better than ½mm or 1/50th inch.
For a ½mm change in a fret spacing of 30mm there is a pitch change of 1½cents. This is good
tuning, and various effects from real strings and frets can cause greater, but still acceptable,
pitch changes. Modern measurements of fret positions to 0.1mm can reassure us that good care
has been taken, but in practice the nearest ½mm may be adequate. Slightly better accuracy may
be achieved in adjusting lute frets, but this is less easy with fixed frets. An ancient maker would
learn from experience, and fine-tuning would depend on adjusting heights of action, frets,
scallops and fret wire.
For a 1mm error the pitch change is about 3cents, which is still not disastrous, relative to the
effects of stretch- sharpening. It would be particularly significant how these errors fall between
neighbouring frets, strings and notes. The 1½mm deviations on Palmer’s frets 5 and 7 initially
appear more important than on Rose’s single fret 3. If these simple basic frets are so far from
correct positions the whole system might be in question. However, consideration of various
transformations removed this possibility and just leaves a straightforward tuning error of about
5cents.
With the simpler frets 5 and 7 one is curious how errors occurred, but a fret 3 requiring a
difficult construction and more adjustment is understandable. The flatness of Palmer’s fret 5
could be a simple error with the saw, a distraction in the process of marking out etc. It may have
gone unnoticed because fret 4 in QCM is also flat, so that the spacing 4/5 is not small. The
deficit appears in the preceding small 3/4. His skewed fret 7 may have been an attempt to
recover a better position for the treble side, but other scenarios are possible. The visually
complex pattern of SCM frets might be an important factor in overlooking errors. The simpler
features of ET and ECM could help to reveal mistakes. The added complication of slanted frets
greatly increases the problem of recognition.
From these considerations it is likely that Rose’s orpharion was played, survived as an exquisite
artefact with historical associations in a noble family. Palmer’s orpharion may have been
marginally playable, but there may be no clear fingermarks. It could have been left in a
proverbial cupboard, and as time passed it remained unworn, decoratively preserved, and worth
keeping.
As a whole, the citterns probably span this range. Of the few examined here, none appear so
good as Rose’s or as questionable as Palmer’s. It would be unduly pessimistic to conclude that
all three dozen were a self-selected set of duds. Evidence has usually survived for peculiar
reasons that can make it seem untypical, and one just has to make the best of it, but not ignore it.
(Much surviving Greek pottery was found not in Athens, but preserved in distant Etruscan
tombs, and papyri of Greek texts were kept dry in Egyptian rubbish tips.) The cittern frets
examined here do have a similar ‘look’, taking into account the above variations and sources of
error. They all display the same idea of tuning, and it is almost conceivable they could derive
from a ‘single strip’. Any possible flavours from different regions, times and types of music
might be difficult to separate from other variability.
Whatever the accuracy of tuning in all these instruments, the present analysis shows without
doubt the intended temperaments. The characteristic patterns of the frets are an excellent guide,
even though some frets are not accurate. This supports and extends previous comparisons with
strips of fret positions. Patterns for ET and for the strong tempering of SCM are well known,
and the analysis has found several new distinctive patterns, such as 1/8th, 1/9th and 1/15th comma
meantones.
FINAL REMARKS
A central interest in these papers has been exploring the technical nature of the ancient approach
to tuning and its results. Conclusions about the ancient methods, together with this later
examination of real frets, have indicated the original development of temperaments for fretted
instruments. Some observations on current and previous modern views of temperaments may be
helpful. Then a final summary of the tunings on the various instruments will round off.
At the end of this study a glance into the web gave two immediate references to ‘1/6th comma
meantone’. The lute-maker Ray Nurse uses it most of the time because it simply sounds better.
In a 2007 recording of Purcell songs, Carolyn Sampson is accompanied by harpsichord, archlute
and theorbo. The keyboard is listed as 1/6th comma so presumably the lutes were similar, like
the ‘quick-change’ theorbo cited above. This illustrates how performers and makers are voting
with their ears, and even departing from ET for music written after it had become more
common. This is a swing from views and practice at the beginning of the early music revival.
(Hopefully the recent owner of a new bandora in QCM will not be going any further.)
It is worth outlining the position several decades ago. First it is necessary to discuss ancient and
modern classifications of tempering and equal semitones. For our modern analysis we can use
exact narrow definitions, but the ancients could not think this way. Terms such as weak, light,
medium and strong temperament or WT, LT, MT and ST might be better for ET, ECM, SCM
and QCM. More likely, their ranges were broader and fewer, such as weak WT for our ET
(1/11th comma) to 1/7th comma, and strong ST for perhaps our 1/7th to 1/5th comma. Our ET
would actually be an extreme, not very far from untempered PYT. Straying towards PYT would
not be helpful unless ‘sharp’ alternatives were used to approximate tempered flats. The other
extreme was QCM with the perfect major third, but difficult fret spacings. This may suggest
1/7th comma as a more practical midpoint than the ideal SCM.
The valuable musical surveys of tuning systems made by Barbour and Lindley provide concrete
written examples. Barbour assumed ET was standard for lutes from the early 16th century, but
specific instruments were a small aspect of his enormous scope (Ref 7). Much later, authorities
such as Praetorius (1618) and Mersenne (1636), indicated that ET was becoming widespread.
Interestingly, this first half of the century is roughly when I concluded that ET would become
necessary for pieces in more distant keys on the renaissance lute and especially the D minor lute,
while the surviving pieces for transitional tunings could continue to use ‘meantone’. Some
earlier theorists such as Vicentino (1555), Salinas (1571) and Zarlino were also cited in Ref 7 as
writing that fretted instruments were always in equal temperament. This is clearly untrue for
citterns, orpharions and bandoras. The apparent contradiction with most of the written fretting
schemes for lutes can be removed by noting that the temperings for lutes were constrained to be
weak, and below about 1/6th comma. In contrast, the early keyboard schemes devised by the
above three writers had very strong tempering that was effectively 2/7th and even 1/3rd comma,
which appears unusable for any music. Lute players might have considered they were using
quite even semitones, although not our exact modern ET, and keyboard players would have
known their own semitones were much more uneven. As Vicentino wrote, they could not play
together, and there are few contexts or surviving pieces where this occurs. Ancient writers were
not endorsing exact ET, but rather commenting on the lute’s technical need to use lower
tempering, which in turn could lead to more equal semitones for convenience. Careful
distinctions need to be made between what was ideal for the music; technically possible and best
practice for the lute; compatibility with keyboards etc; and possibly a useful shortcut. Any
ancient use of similar semitones for convenience would be similar to exact ET in the modern
revival, where our wide range of styles makes it understandable. The written sources have often
been misconstrued as evidence that an exact ET on lutes had been a preferred ideal standard for
much of the 16th century. This view was combined with uncertain evidence in paintings, and the
encouraging fact that ET had become modern practice. Problems have arisen when any
indication of early usage of similar semitones is mistaken as evidence of a musical preference
for an exact ET, rather than purer tunings, throughout the whole early period. The matter can be
reconciled by the present explanation of different notions of tempering and ET. Over time it
appears that tempering decreased in lutes and in keyboards, the former always being generally
less tempered. Finally, after the lute disappeared, keyboards using strings also became ET.
Later, Ref 9 started from the assumption that equal temperament lute frets were standard from
1550 to 1650 (p19) and aimed to find an even earlier date. It was considered that players might
bend notes to match the natural tuning of singers, flexible instruments, and even strongly
tempered keyboards. In practice there is no way of flattening sharp frets in ET, and the converse
is not consistently easy. The lute seemed to have been granted the status of a modern piano to
which other instruments needed to adjust themselves, while the solo lute would also be content
with an unnecessarily compromised tuning. The final conclusion that ET on a fretted instrument
can ‘never be historically wrong’ (p93), perhaps hinting ‘always historically right and
compulsory’, was unhelpful even before the present examination of real fixed frets. This seems
unusual when it was also noted that French harpsichord tempering was probably changing from
around SCM to ET over the early 17th century, which coincides well with the new conclusions
on lutes. (The following musicological excursion of three paragraphs could be skipped.)
In Louis Couperin’s pavane in F# minor, Ref 9 suggested several musical effects for a possible
tempering around 1/5th comma. Modern keyboard players think the key is ‘bizarre’, and in
Couperin’s total of about 130 pieces there is only this single piece in F# minor, four in A major,
three in B minor, and four in E minor. However, lute players know that he was using keys
explored by lutenists, who had originated the new musical style. In addition the dance form was
an old type but still used by lutenists. The piece can also be seen as an unascribed lament or
tombeau (tombstone), which became a frequent dedication from lutenists and harpsichordists,
but was more usually like the allemande that succeeded the pavane.
Couperin’s pavane may have more detailed interest, at least for us if not originally intended.
Dufaut wrote two long sombre pieces in remote sharp minor keys: the pavane in E minor and the
allemande ‘L’Offrande’ in F# minor. These might also be unascribed tombeaux, and the latter is
simply ‘the offering or gift’. Other lutenists such as the Gautiers wrote pavanes and tombeaux,
which were generally shorter and with clear dedications, in keys up to F# minor. The above
analysis indicated that the D minor lute had recently needed to use ET, because of the limited
consistencies in meantone. Harpsichordists regarded their instrument as an enormous loud lute,
and they could also use more general temperings rather than meantones. Couperin’s piece in a
full 4 or 5 parts therefore has aspects of a grand demonstration of size, volume, tunings and key,
and perhaps a tombeau for the lute and its old tunings. More subtly, however, it was found
above that many lute pieces in remote keys could actually be played in a special ‘sharp’
meantone. If they had known this, lutenists might have pointed out that they had a more decisive
tuning than a keyboard, in addition to a better tone. It may not always be clear whether remote
keys on a keyboard should exploit peculiarities in tunings based on more central keys, or
involve changes to suit the new key, which would be an unusual step for this single pavane.
However, all this technical detail is exceeded by the emotional power of Couperin’s piece,
which is a raw intense grief. He left only one other tombeau, which is also like a pavane but in
the contrasting key of F major, and this has the dignified gravity usually found in tombeaux. It
was dedicated to the lutenist de Blanrocher (Charles Fleury), who remarkably received three
other restrained tombeaux: by Dufaut in G minor, Denis Gautier in A major, and Froberger in C
minor. Couperin had moved to Paris in 1651 where he met Froberger, then de Blanrocher died
in 1652, and a year later Couperin’s father also died. Perhaps the F# minor pavane was for his
father, a companion piece for de Blanrocher, for someone else, part of some comradely or
competitive writing in this new tombeau genre, or some combined influence in this short period
of excitement and distress just before he took appointments as organist and violist. A date of
composition would be useful, but may not be obtainable from the pieces copied out after
Couperin’s death. On the difficult question of dates, the composition of tombeaux might be
bracketed more reliably than other pieces, simply from the two deaths in the absence of other
information such as publications. For example, Ennemond Gautier probably wrote his famous
tombeau for Mesangeau shortly after 1638, unless he had it ‘in a draw’, and definitely before
1651. Interestingly, Dufaut’s tombeau in G minor, his E minor pavane, and ‘L’Offrande’,
together with only a prelude in B minor, are found in a single source, the important Saizenay
manuscript with the late date of 1699. All these links appear to centre on de Blanrocher, and his
many dedications may have some grim relation to a fatal fall (tombée) down stairs. It is of
course not necessary for a lament to have a specific subject, and Bittner later wrote a piece in F#
minor simply titled ‘Tombeau’. Also, it is well known that many French pieces have titles with
no clear meaning, and dedications to people now unknown.
After this detour we can return to the main subject. The idea in Refs 7 and 9 that equal
semitones and formulae for ET were a constant goal for the ancients tinged the brief survey in
Ref 1, which has been left unchanged. A similar non-sequitor would be noting that a small
proportion of Greek poetry had used the iambus, and then concluding this was their standard
because something with the same name was used later in England for several centuries.
However, we are all constantly making false assumptions, and holding incompatible ideas. For
the last century, exact harmonics and difference tones have sat happily with ET and cents. In
contrast, the ancients had more inharmonicity in their instruments, but a greater use of more
perfect musical tuning. The present study of evolving ideas, ancient geometric methods, and real
frets shows that the previous view of ET was too influenced by simple modern definitions and
writing history backwards, but it seemed self-evident thirty years ago. Any future book on
tuning and temperaments would probably need several authors with special strengths in history,
maths, music and science.
For fretted instruments throughout the 16th and early 17th centuries there would have been a
continual use of purely Pythagorean intervals, an inclusion of other perfect intervals, and the
development of geometric constructions from these intervals, all supplemented by finer aural
adjustments. A main aim would have been to describe a generally tempered tuning as much as
any specific degree of temper. This broader aim was determined by the available mathematical
methods, with no early explicit division of a comma. Precise systems and labels in modern use,
such as ‘sixth comma meantone’ and ‘equal temperament’ were unknown. Many of the
schemes, including Gerle’s, would have been used as an initial start for a range of tempering.
A related, but seemingly different, aspect is the new finding that SCM is closely related to the
perfect intervals favoured by the ancients. The links are rigorously theoretical and also practical,
showing that SCM is in fact a special case, rather than a random choice with modern popularity.
This is useful because modern SCM has almost come to seem as singular as ET, but more
arbitrary. It may seem odd that page 1 of Ref 2 began with a list of perfect intervals and we have
returned there many pages later. As someone else put it ‘We shall not cease from exploration,
and the end of all our exploring will be to arrive where we started and know the place for the
first time.’ However, the initial interest was an exact consistent system. The study of old
schemes then found that the unusual fractions and interpolations were attempts at perfect
intervals and tempered modifications. These were scattered and not completely consistent with a
definite meantone, and the present study of real frets shows similar variations. A use of perfect
intervals in the ancient theoretical approach to tempered tuning, and also in all practical aural
methods, may form a satisfying conclusion to the inquiries.
A thought experiment could illustrate the relation between aural and theoretical methods, and
conditioning to current ideas. A player in ancient times tuning aurally from perfect intervals
would most likely be near SCM, or tend to QCM if many intervals were sacrificed for perfect
thirds. A modern player if asked to tune open strings and frets ‘from scratch’ is likely be near
ET, but if asked to repeat using perfect intervals would probably produce an ancient type of
tuning. A panel of perhaps four real modern players might be interesting.
For a renaissance lute, the range of tunings would probably have been between about 1/5th
comma and ET, depending on time, place, type of music etc. It is possible that lutes had some
tendency to the cittern’s good chords on low frets and more sketchy high frets. There is an
indication of this in the schemes, but the more contrapuntal music suggests a more even tuning.
Later, a more definite ET would have become useful for a wider range of keys on renaissance
lutes. For the new transitional tunings probably all surviving pieces could have continued to use
a general meantone. The final D minor tuning on baroque lutes would have needed ET for most
pieces, but several special meantones were also possible. In general, greater tempering may have
continued in contexts where its better tuning was allowable.
There is some similarity between our precise modern ET and our precision uniform lute strings.
The ancients had a range of non-uniform tapered strings, resulting from the natural variations in
guts. There would be a low chance of uniform strings, in the modern sense, and the ancients
would have no practical concept of such precision. Similarly, they could have no practical feel
for an exact ET until the later irrational (2)1/12, or perhaps the 18/17 rule. A range of tempering
would be analogous to the natural range of tapered strings.
Wire-strung instruments had tuning close to SCM, although some upper frets had noticeable
tendencies to ET. The citterns had lower frets that were sometimes more tempered, reaching to
QCM. They could be closer to a perfect tuning than lutes, which needed to compromise because
of the number, position and tuning of open strings and more adventurous keys. Rose’s orpharion
was very close to an exact ECM, and there may be some relation to the lute. He was primarily
famous for making viols, and the tempering on his orpharion may reflect a similar approach for
viols and lutes. If one expected an evolution from Rose’s ECM in 1580 towards ET, then
Palmer’s 1/7th comma 37 years later would be a great surprise. Palmer was also a viol maker.
Likewise, citterns continued with their SCM well into the century, and presumably so did their
fellow bandoras. An orpharion in Frankfurt has short string lengths of 412 to 484mm. It is
attributed to the viol maker Tielke and probably dates from as late as 1680. The frets are said to
be even later alterations (Ref 23). Important knowledge may have been lost for an original
tuning used at a late date, close to Vermeer’s painting. The other orpharion, in Braunschweig, is
said to be damaged and poorly rebuilt. Both may be worth pursuing for signs of their original
frets.
The brief lifetime of the orpharion coincided with the golden age of the English lute. It would be
easy to assert that the frets of two orpharions were representative of all lutes, but any relation is
strongly indicative rather than proven. The orpharion was promoted by William Barley’s
collections, and Dowland was annoyed by this diversion from his own publications (eg Ref 3). It
would have been interesting to have Dowland’s musical and technical views on the orpharion
itself. The restrained playing dictated by a low action and tension produces a reticent sound, but
more freedom seems possible on citterns. A wire string produces a long sustained sound in
contrast with a shorter decay from gut. With modern overwound bass strings the prolonged
notes are often a musical problem, and the attitude of the ancients to the harplike sustained
sound of an orpharion might have been similar. For a harpsichord, with a large resonator,
plucking by hand produces a marvellous sound. The nearest special stop might be ‘peau de
buffle’ using plectra of buffalo skin, but this cannot emulate a large soft fingertip. (The ‘lute’
stop simply places the normal quills closer to the nut - and the cittern was often played with a
quill. The ‘buff or harp’ stop places a damper over the nut.) It is rare to hear special stops that
are nicely adjusted, and it is usually a relief to return to the main registers. Similarly, the
orpharion can be an interesting contrast to a lute, or a substitute. Its key practical advantage was
that wire strings would be very uniform. In the process of drawing through a hole, any non-
uniformity in the plastic flow quickly becomes an instability and the process breaks down. Wire
strings with non-uniformity down the length are a virtual impossibility. This is very different
from a natural gut with a continuous variability over its entire length. The many obscure
writings on gut strings would have been a response to the problems faced by amateurs taking up
the lute. If players gave up then sales of music books would suffer.
This unexpected renewal of interest in temperaments arose from attempting to understand
ancient instructions for fretting and tuning. Future tuning topics might be combining ancient
fretting schemes with non-uniform strings of limited elasticity, leading to averaged fret
positions; and proofs concerning just intonation. The properties of ancient gut strings were
discussed in Refs 16 and 18, and a shorter following paper extends this work. The search for
improved sonority in baroque lutes, together with the different tunings and the music, may be
helpful for understanding old lutes and their strings. The baroque developments appear less
meaningful in relation to modern strings and lutes.
A little spare time at the end of this study has allowed me to set up a lute with transitional
tunings and a general meantone – 1/8th then 1/6th comma. The result is very pleasing, especially
for the full chords using open strings and major thirds in the music’s narrow range of keys. This
original use avoids the lower tuning accuracy of ET, which was intended for a large number of
keys that are not used by the music. Any small changes to higher frets needed for a few pieces
would be easier than retuning one or two courses for a different ‘accord’. The tuning meter
settings given above can be adapted, depending on a preferred reference point, such as A for a
3rd course in Mesangeau tuning or for a 6th course instead of G.
Just after the present work, a meeting on ‘accords nouveaux’ concentrated on performances, and
classifications of the many historic systems. A total of 175 open tunings appear in sources, but
only the handful analysed above left much important music. This large number is
understandable with 3 possible pitches for each of the upper 3 courses, and 2 choices for the 3
lower courses. This produces 216 possible tunings, which is a better guide than say 3 pitches on
6 courses and 729 cases, or 8 courses with over 6000 cases.

FOOTNOTE on CITTERNS
A fortnight before I was due to submit this paper, Peter Forrester kindly offered his tracing of
frets on the earliest known cittern, made by Plebanus in about 1536. This arrived with two more
tracings of citterns, by Virchi (formerly Stradivarius) around 1575, and Salvatori around 1600,
all now in the Musée de la Musique in Paris. An initial thought that the previous work might be
invalidated turned to an impression that the earlier conclusions might have been reached more
easily with the new data. A considered view is that there is good consistency with the previous
findings, and some interesting new features. Plots of the fret positions on the diagram show that
all three citterns have frets tuned for about SCM or 1/5th comma. The main aspects can be
summarized briefly, but a full presentation of the diagrams and analysis will take time.
All three citterns reach fret 19, and Salvatori’s has all 19 but the other two miss frets 4 and 18.
This highlights the basic relations between frets 7, 12, 19, that were only displayed above by
Campi. The simple geometric proportions of the string are 1/3, 1/2, 2/3. Plebanus shows this
precisely, but the other two have small variations up to about 3mm. It might seem that
allowances for stretch-sharpening are involved, as discussed above, but the variations have
differing directions on the two instruments, and this is examined below.
Another useful feature of all three traces is that they allow a good accuracy of measurement. I
could measure to about the nearest ½ mm, plus or minus ¼ mm, whereas about half of the
earlier sets of data were far less accurate. However, some of Plebanus’s frets appear to have
been reinserted upside down (Ref 6) showing edges of doubled up sheet, and with wedges less
usually in front. The analysis assumes that the tracings are best judgements, but further
assessment would be useful since this cittern has unusual details.
The important lower frets 1 to 7 are around SCM for all the citterns. This is particularly
significant for Plebanus’s tuning and frets, which were most likely established before Gerle’s
particular writing. Several other possible earlier theoretical influences were discussed above. For
example, citterns may have helped Gerle, who in turn later helped cittern makers, etc.
Plebanus’s higher frets 7 to 11 are also approximately in SCM with its characteristic alternate
spacings. These frets may be even more remarkable since they were not included by Gerle, and
it was found that later schemes had theoretical difficulties with upper frets. Plebanus probably
made fine adjustments to fifths from lower frets, filling in, etc, as discussed above.
In greater detail, for Virchi and Salvatori the important frets 1 and 3 are more tempered at about
1/5th comma. This was seen on other citterns for fret 1, but less clearly for fret 3, while Campi
was exactly SCM. Surprisingly, Plebanus has a less tempered fret 1, in about ECM rather than
the common tendency towards QCM. Again, it may be significant that this early cittern was
made decades before the others, and also before ideas and schemes such Gerle’s with the precise
frets 1 and 3 may have become well-known. It was noted above that a flatter fret 1 nearer the nut
would have more stretch-sharpening, especially for a greater nut height, and this would be a less
stable method to obtain SCM. In contrast, Plebanus has an upper octave fret 13 that is sharper
and distinctly QCM. Clearly, this fret could not have any extra sharpening from the nut. Taken
together frets 1 and 13 may indicate some ancient uncertainty in construction, or simple error,
but this is a contrast with most of the lower octave frets which are close to SCM. This very early
cittern may be an isolated case. Further information would be useful on the probable action
height, scallops etc, in addition to the replaced frets.
For the citterns in the main text, apparently random or scattered patterns of fret spacings were
noted, particularly for frets 7 to 12. A major effect of the higher accuracy is to confirm this as
real, but the scatter is a little lower as for the two orpharions. The general impression is that the
ancients thought slightly random looking spacings were an acceptable result of their whole
process of theory, marking, making, and playing. In an unusual contrast, Virchi’s upper frets
have an almost perfect gradation of pairs of equal spacings. It is likely that these were sketched
in to fill the space 7 to 12, as suggested above. This might be part of a theoretical construction
with adequate simplicity, or tidying up possible errors, uncertainties and the appearance.
For the previous citterns and orpharions, the upper octave frets were a reasonable match with the
important lower frets. Both octaves were close to SCM, and the upper octave might have been
constructed from the lower frets. An unexpected feature of these three citterns appears to be
more scattering on the diagram for the higher octave, whereas the previous citterns had more
scattered frets 7 to 12. These effects are not greatly changed by reasonable transformations, and
the comparison may be just chance with all these relatively sketchy, but probably musically
acceptable, less crucial higher frets.
Above Plebanus’s fret 13 is a small spacing to fret 14 then three equal spacings to fret 17. A
small flattening of about 2mm on fret 13 would put it near SCM with four equal spacings. This
is a good illustration of how possible errors in initial constructions, marking out, sawing and
fitting frets could lead to apparently random spacings, and also how a single small change can
greatly alter the pattern. This would be more likely for higher frets, where the size of the random
errors of order ½ to 2mm is comparable with the characteristic differences between the alternate
spacings in an exact tuning such as SCM or ET. In contrast, the errors are smaller than the
differences of order 4mm that distinguish tunings of the lower frets, which consistently display a
strong tempering. An opposite process of tidying up the pattern of frets seems to be far more
rare. Small variations could produce four types of combination, with regular or disordered
patterns each leading to good or bad tuning.
For the four citterns extending to fret 19, some important detail is now possible. An important
comparison can be seen by taking fret 7 as a nominal 200mm from the nut, and listing the
relative values for frets 12 and 19 as:
 Plebanus 299.7 398.9; Virchi 300.7 398.0; Salvatori 303.4 403.1.
Campi 303.2 401.4; SCM 301.2 401.2; ET 300.7 400.7.
This may be clearer than taking the more obvious fret 12 as a reference point. Close examination
shows that none of the citterns follows the same pattern as another, nor an exact SCM or ET.
This indicates that there is no possible universal correction that can normalize the positions. Any
accurate tuning would generally have required small individual corrections of scallops, action or
playing technique. Alternatively, tuning errors of order 1mm in 200, equivalent to 8cents, may
have been considered acceptable. If fret 19 was used only rarely, and a principal purpose was
setting up the frets, then the error would be smaller. This should be valuable understanding for
modern makers taking a historical perspective, and may be as significant as long lists of exact or
approximate temperings. A further stage of analysis could benefit from detailed calculations of
stretch-sharpening, using best measurements or estimates of nuts, scallops etc on the old
citterns. Calculations based on a finished modern instrument would also be useful for
understanding, calibration or future design.
An average cittern can be defined from all the scaled frets examined above:
Av: 36½, 64, 97½, 122½, 150½, 177, 199½, 222½, 242½, 264, 282½, 300, 318, 332½ 348, 360½, 374½, 387, 398.
SCM: 36, 64½, 96½, 122, 151, 178, 199, 223½, 242 , 264, 281, 300, 318, 332, 348½ , 361, 375½, 389, 399½.
ET: 33½, 65, 95½, 124, 150½, 175½, 199½, 222, 243½, 263½, 282, 300, 316½, 332½, 347½, 362, 375, 388, 399½.
All nine instruments have been used, including the two orpharions, and the two citterns with
poor data. This does not greatly alter the result for the five good citterns. Deviations of 1mm or
more from an exact SCM have been underlined on frets 3, 6, 8, 11, 17, 18, 19. Most of these are
not too important, except for a sharp fret 3, in contrast with the problems of flatness for Rose’s
fret and some of the schemes. If the large value for the Dutch cittern is left out, the average is
reduced to 96¾mm. Returning to the photo, fret 3 looks ‘normal’ and I made an early error of a
division – an authentic illustration of a likely ancient problem. Exact ET values are given for
comparison, and the underlines indicate the few cases where ET is slightly closer than SCM.
The effect on frets 5 and 7 is not significant, for reasons explained above. An eventual average
from all surviving instruments might be called the ‘Cittern of Citterns’, with the attractions and
weaknesses of such averages.
An actual tracing shows the frets as short dashes on a single line, with the three string lengths
side by side on the same sheet. The eye easily picks up several general features of similarity and
difference. Three tracings seem a useful number and many more on the same sheet could be
confusing. A missed fret reduces the potential information, but the loss of two spacings and a
continuous pattern seems more of a problem, especially around fret 4. There may be some
relation to difficulties with this fret in later lute schemes. With several higher frets, and gaps for
frets 4 and 18, the spacings around fret 12 attract attention. The theoretical spacing 12/13 is
never equal to the previous 11/12, which is always greater by 1 to 1½mm, for all scaled frets
between PYT and TCM. Recalling the equal pair 10/11 and 11/12 found at 1/9th comma, an
almost equal triplet occurs between frets 10 and 13. At a slightly higher temper the pattern tends
to S L S, and real frets with three equal spacings could be almost perfectly tuned. This can be
seen for Virchi and Salvatori, with even longer sequences of almost equal spacings. For
completeness, a similar triplet occurs near 1/15th comma between frets 11 and 14. Finally, we
may wonder at these ancient fixed frets, similar to the feats of the original geographic explorers.

FOOTNOTE on Pierre Gaultier

After submitting this paper I have seen the CNRS edition of Pierre Gaultier’s lute works. It may
be unfortunate if his simulation of canon fire and bells in only 3 pieces has influenced his
modern reception more than his other 100 excellent pieces. Even the very greatest composers
had difficulty surviving their battle pieces.
Many of his pieces are for the transitional tunings and keys examined in the main text, and
would be playable in a general meantone. In addition there are pieces for French Flat tuning in
the keys of Eb and Db major, and Bb minor, which were not seen for the other composers. These
are either side of the usual Ab major and F minor. The analysis in the main text indicated the
further keys could be compatible with a meantone. However, there are 47 pieces by Gaultier in
the additional keys, far more than in the usual keys, and a full treatment would be lengthy.
Pierre Gaultier also used a variation on French Flat tuning in which the 6th course G is lowered
to Gb. This course therefore changes from the ‘sharpest’ to the ‘flattest’, and the open tunings
are spanned by 6 fifths from Gb to C, rather than 5 from to Ab to G. This produces fewer
common enharmonics, so that compositional possibilities are more limited, similar to
Mesangeau’s tuning. Gaultier’s 5 pieces for this tuning are in Eb minor and a detailed
examination shows that a general meantone would be possible. An exception is fret 1 on the 3rd
course, producing a Bbb that is sharper than a true A. This occurs only once in two pieces, and in
a small group of three in another piece. However, some of his pieces, such as the sarabandes, are
so beautiful that a single exposed wrong enharmonic could need attention, perhaps by nudging a
fret or using a tastino. In a single case fret 6 might produce Ebb rather than a D, and the ancient
compromise position may be satisfactory. These pieces are a technical feat in using mainly flats,
with little use of the naturals, and avoiding double flats.
At the same time Pierre Gaultier’s musical style may suggest the work of later composers using
D minor tuning, more than other pieces using transitional tunings. He left no surviving pieces
for vieil ton, and only 6 of his pieces use D minor tuning. These are in the key of D minor and
could use a general meantone. However, there are a few instances of F# on fret 1 and G# on fret
3, and the main text indicated an early pressure on the new D minor tuning to use ET. This
additional work has been useful in showing that Pierre Gaultier stretched the possibilities of
meantone and keys beyond his contemporaries, but he did not make a large jump to keys and
fingerings that would only have been possible in ET.
It is worth noting some differences between meantone on lutes compared with keyboards. As
seen above, lutes can use only a limited range of keys, but can have alternative enharmonics on
different courses. In contrast, keyboards are not constrained to a simple meantone, but do not
have any alternative enharmonics, unless there are split keys. Keyboard tunings can be
acceptable and provide distinctive features for a wider range of keys. However, some extreme
keys can produce problems of sourness, without appropriate expressive effects. As an example,
for Louis Couperin’s pieces in C minor the Ab can be too flat when G# is the preferred tuning,
and in F# minor the E# can be too sharp. With only a few pieces in these keys it may be
interesting to retain the same tuning used for more central keys. To obtain greater euphony,
changing a single enharmonic would be easier than altering the central point or the overall
degree of tempering. For lutes, in the same spirit of economy, the same basic meantone and fret
positions need only one or two changes, even for the special ‘sharp’ meantone discussed for D
minor tuning.

REFERENCES
1. C. J. Coakley: FoMRHI Comm ‘Dowland’s Lute Tuning’
2. C. J. Coakley: FoMRHI Comm 1808.
3. I. Harwood: Lute Society Booklet No 10, 2005.
4. D. Gill: Lute Society Booklet No 3, 1977.
5. The Renaissance Cittern website.
6. P. S. Forrester: personal communications.
7. J. M. Barbour: Tuning and Temperament, a Historical Survey. 1951.
8. E. M. Dombois: The Lute. Vol 22 Part 1, 1982. Also JLSA,vi, 30-32, 1973; and vii, 1974.
9. M. Lindley: Lutes, viols and temperaments. 1984.
10. P. S. Forrester: The Lute. Vol 23 Part 1, 1983.
11. P. S. Forrester: Lute News 75, 2005.
12. C. J. Coakley: FoMRHI Comm 1809.
13. D. Gill: The Lute Society Journal. Vol 2, 1960.
14. D. Abbott and E. Segerman: FoMRHI Comm 124.
15. Ashmolean Museum website.
16. C. J. Coakley: FoMRHI Comm 1810.
17. C. Goodwin: Lute News 57, 2001.
18. C. J. Coakley: FoMRHI Comm 1812 & 1814.
19. Thomas Mace: Music’s Monument. 1676.
20. Centre National de la Recherche Scientifique, various complete works.
21. C. J. Coakley: FoMRHI Comm 1811.
22. C. J. Coakley: FoMRHI Comm 1896.
23. S. Barber: Orpharion string questions- on web. 1998.

KEY for DIAGRAMS

Rose’s orpharion: Circles with crosses are frets 1 to 12.
Squares on the first three frets are equivalent values for frets 13 to 15.
Squares in brackets represent perfect fifths on frets 1 to 4.
Letter P indicates perfect intervals.
Solid dots represent a shift of the nut 2.0mm away from the frets.
English cittern: As for Rose, except no solid dots, and
Squares on the first five frets are equivalent values for frets 13 to 17.
Campi’s cittern: As for Rose and English cittern , except no perfect fifths, and
Square in brackets represents a perfect fifth below fret 11.
Solid dots represent a shift of the nut 1.5mm away from the frets.
Brussels cittern: Circles with crosses frets 1 to 12. Squares on first seven frets for 13 to 19.
Ashmolean cittern: Plain crosses frets 1 to 12. Plain circles on first seven frets for 13 to 19.
Palmer’s orpharion: Plain crosses are frets 1 to 12 on the treble side.
Plain circles are frets 1 to 12 on the bass side.
(as Rose) Circles with crosses are the average of bass and treble for frets 1 to 12.
(as Rose) Squares on the first three frets are equivalent values for frets 13 to 15.
(as Campi) Solid dots represent a shift of the nut 1.5mm away from the averaged frets.
Dots in brackets represent a shift of the nut 0.75mm away from the frets.
Crosses in brackets indicate a change in fret 12 on the treble side.
No perfect fifths on frets 1 to 4.
Geometry of string stretching: Scales highly distorted to show important features. Lengths and
heights explained further in text.
The diagrams contain an enormous amount of information. The fret positions for an instrument
and a chosen condition can be followed by selecting a symbol and tracing across the frets with a
fingertip. For greater clarity, enlarged A4 copies could be made and emphasized by coloured
pens.
 *t)
-raa

ffi
*d*
n{t*

=-\
LQ
il-_
S_* =rffi b*<E
".&b
-ffi
"'ffio - %e*r }Fgib:qE Eaafi,gg

tii'

-.8
_ _.- *"lL_
-"tt-
\t*
4-& w ---re* ffi".
I{tl*h*ia'
b, E-
.M.
\ce#d
*

#q
W-"

*
"\,\L
'E - --
I rqfL

k-
_. ffi ffi*
ft
-rr- _. *__ -t_ q'
-r.rrr
_t_t <_r*
- r
-- - q,sr
ull
f,t
frt

Hi
El ry-
r{I
*- rb #P't
fll.e.-
dlt
sl
r{I
f-fi
HI
ol
#l fl%
€l
ril
rqI
rar.. _ *l .*L
trr'&s

cl -:_=:.#
-{}'
ql
cl Eb
kE _-_i :t_
_ hL
'*-t=
'o{ r rll l+-
+rl.
o _{l
ral
cl
s4
EI
rF
H{l ll
TL".
€*a
Wk % w

II
r-{l
.
5t
-{l
L4l
cgl
ril
El
ol
3+J

al
<nl
(}I
cnl
&
OI
pcI

=.i_.. -_s-
.dF*-

all\
l-r ,,
\J i
ta,,

-r-*
\
-_
.-i_

__- t
-r q
ffi
% -&
&
Fh#&eB

-
_ 6k

_ _._
$* --_kg t *rrL ,<E+
_ql *qr-

-- c_
qqr 4
rrl
{-
t-{ fr)
OG
'rr \{}
ar-
h-{ tn
\<5
__-__ +"
-8\r/1
Ct crl Ct
fi.
ct rt l* ft il.
vj ':
4.,-
11 \J
H lr'?
I*t ?-{
L-r'
U
01
rrl E
,U
H II I} FN
rt
Ii F*
rt'l flfi afr ,t-;
{i'-i
*"q#
 i

@'-
r;& G # ## '@ *\ -,t

4
€
"E3iE s \

\#ft *-.r.8
b

---*\
l;L*x
l-
to
_.*,

_E
_.-
q

--..
iaa,
w .kr,
_ t!, ._rEg
_qc- ,sr
r-t*
*-i?-
--d;-
'
"\Lb

-\
4q' ,
'aa *.^.".-fu-.,

ffiffik -
-
tr nE

-ffip{W ffi
JtGk4&x

i-r T-_ ------:

,*$&
.L
#
aE* &"
&,r_&^j; \**/
eeffi.,&

+Ltr
<ki"
__n+ .k
rt r _rF
..-qsi
-ae. _b:

u)l
{rl
ci
Hl
dl
i..{ I
'#
c)l
al
EI
(.)I
*Jl
-ol .* _q+

gl *&
rr-a- -*oh

X
trl
Et?
!-!-
B .'h
% _Eb
rtEE q*
__ .%
-*_.
:Er1
.
ol
r-rl
+rl
ril
o
al
cl
w
A{ &*x.
6l #,eL
!.

,.!l @* fu&h
tl
trl
I.<
()l I

+-r I
+Jl
ul
tu
. F{l

t#
-ql
o
a1
ffi
m
-<lI
r-r
a0 &
Bi

sl
trl I

.s\
t-r
ru _t-&
.-q
.ErE

-., -ai_ =*_ rt-r: ,&"

q
.:!.

d{h
ffi ,*"&
&
!& tu-
ldq*6+sEt

*P' o&
n -t '
ant' -"*
'*

--b-
. tru%

*Q- ''{k.

$--} -.q

Ct
E*
fr]
CI*
-qr \*
rt
H
Z
I',fa
r-'
' +n-- .-.%€!
*1
*w
q*r-
crl
ct
f,n
11 t* ! f-r UH U
Rn l'r''l a'-- A\J
r*t q a
Ii rl
it
!t
ft ,I
I'l
U
H
L+
t*
ff/.1 ff1 *
Fft
t"*
 *.L}
*q.
r\ t

.--re
,\,LC
-rEF

al
r-r I

trl
Hl '*
F{l
dl
-€ffi-
bl
pl @
EI
()I
Pl
'.(Jl
ql
LUI
"<h +

-l
trl
'qEY

.
ot
-tl
-lJl
. F{l
al
ol
pl
6l #ffitt
,.Hl
rl
4,
-il
F{t
Ltl
OI
+)l
-F)l
ul utl
ililt
Al
.r-rI
al
EI
dt
UI

*ffi**iq
EE"*.S U*3

*"4h".
JT
rfi
rr! \r
ffi L-, {r} -='lt+
# '&1 .-

E* t*.- P.-*
s*t
re ra r* w
ft tfli
r"t rt rt .A\
I.J
F4 fr.
\.#

rn
I a
li C)
1i ll il
rFs4"

a
F.
nfr ia*{

*. I
'*%d
 &
@
* .^-h{.*

+Jl
anl -\
HI
€)t
r-l l
HI
dI
f-{l
(],
EI
$l
#l
trI
F{t ** qs f*.,wr
tud
ffi
r3I @ t#t-
*Tp
fil
rdl
vll
#l
(}l
. r-rl
r
{-}l
-ll
.-{L
ml
CI qlF
st
{-}l
all
.L{l
Hl
ti
rJRl
/.{l
ra*
ryii
.T*q-
il(Jl E'I
w
+.rl
.i-l I
' -.f l
UI
(dl
*l
0)l
t-tl
ol t-r
u-1
fil
l-{ I
r-{ I
,.4-i I
{at
<l
.LJI
fit
dt
{*d i
ml
dll
ml *.*&,.
#31
rrd I
;rt t.r -._.a.-

L+J
et
l-{ I
-

- .i*
'

*w
rF
ff: *(} ,'k _

*ffl
t+r,.fr
t4rl
ffl 1* f* tEJ
t tr
tfl ff?
r? r*: ft ra
q & k

tsd r.-t tJ & e

H

$t
II fi il
r*{
C/ rt
{f)
t}
rfr r'rn
ti
.fi
a--*-
t*
*flt'ttr
 *t}
-riq
n \hJ'

.*{}u
-L
-,X*r

al
{+l
trt
ol
EI .trF
t

631
t-{ I
OI
o.t
EI
()l
{Jl

"dl
El -*f+,
utl
dl E3r
ol
.r-{l
9l
.r-{l
utl
ol
a+
PI
()t

flel &#€ua- re*'J-:s

ol
.r-ll
t-.t I
dt
-ql 1;!
st fl"U
ol
^l
AI
L{l
c)l
EI
dl
FiI

**eb
r-
ct
ff] os 1. va qF
*{
t-i ctr,.rl!!,efirrr \r I :?$>
=t-, r-" }
:li
v
ZA
\./
'r r\
C.

*-
=iE
lt i
cC-[
=t;
rll
ct-i
!-{

e
fi
\./
C
{F*,
ro-i
ft
F"{
a
U
E*
*o*.- %
k"
Geometrlr of string stretching

Light pressing

t,

Hard pressing